CN106529465B - Causality recognition methods between a kind of pedestrian based on momentum kinetic model - Google Patents
Causality recognition methods between a kind of pedestrian based on momentum kinetic model Download PDFInfo
- Publication number
- CN106529465B CN106529465B CN201610973556.8A CN201610973556A CN106529465B CN 106529465 B CN106529465 B CN 106529465B CN 201610973556 A CN201610973556 A CN 201610973556A CN 106529465 B CN106529465 B CN 106529465B
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- momentum
- target
- causality
- relationship
- 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
- G06V20/40—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
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of causality recognition methods based on momentum kinetic model, according to causality concept, construct momentum dynamic model, it is proposed that the causality method of discrimination based on momentum dynamic model calculates cause and effect value, three kinds of video line human world causality is identified according to cause and effect value range, it overcomes other causality models and is unable to causal defect between quantitative analysis target, realize intelligent monitoring and accurately identified between causal pedestrian.
Description
Technical field
The present invention relates to video analysis and image understanding field more particularly to a kind of target based on momentum kinetic model
Between relation recognition method.
Background technique
In crowd behaviour analysis field, Activity recognition, the correlative studys such as classification and pedestrian's Attribute Recognition are existing to go through for many years
History.Human bodys' response is divided into three parts, i.e., mobile identification, action recognition and Activity recognition.Although the related row of domestic and foreign scholars
The research of people's behavioural analysis is small quality, but is still confined to pedestrian behavior itself at present and goes thinking and problem analysis, right
Rarely has research in the Crack cause of behavior.External psychologist is by the connection induction and conclusion between nature object, and according to one
Set pattern rule classification, forms the theory to causal relation between the things that describes to be widely present in nature.
Causality theories pass through the development of many years, are nowadays divided into three categories not, are psychological model theory, reason respectively
Model theory and mechanical model are theoretical.Psychological model theory describes three kinds of causalities by way of reasoning from logic；Reason
Model theory is then the theory constructed based on causal Bayesian network analysis；Mechanical model theory is based on power kinetic model
It is established, this method passes through Mechanics Vector relationship description three kinds of causalities.It is developed by power kinetic model dynamic
Mechanical model constructs model, identifies causality between things by force vector between analysis target.
However, the causality research that above-mentioned model is done belongs to qualitative analysis.It is difficult to by concept each in model
The limitation accurately expressed under reality, these types of model are not all able to achieve causal quantitative analysis.
Summary of the invention
The present invention proposes causality recognition methods between a kind of pedestrian based on momentum kinetic model, it is therefore an objective to pass through view
Frequency image analysis technology solves the problems, such as in small-scale crowd behaviour analysis field in relation to the causality identification between target.
To achieve the above object, use following technical scheme: causality of the present invention includes leading to relationship, promoting
Into relationship, hinder three kinds of relationships of relationship, cause relationship refer to target pedestrian due to influences of extraneous factor change its initially
The direction of motion；Promotion relationship refers to that the effect of external force plays the role of promotion and final and have not been changed its side to target pedestrian
To；Obstruction relationship refers to influence that the movement of target pedestrian is hindered by extraneous factor but and has not been changed its initial motion direction；
Target pedestrian's initial momentumMomentum of the target pedestrian by other pedestrian's repulsive forcesPredict last momentumPractical end momentum4 elements as target pedestrian；
The recognition methods the following steps are included:
Step 1, video image is obtained, using cam-shift algorithm, obtains the position of each target pedestrian in video image
Set, speed and range information, in conjunction in social force psychological forces model and momentum theorem construct momentum kinetic model；It utilizes
Relationship between 4 elements of target pedestrian constructs identification condition, the causality in the two target line human world is identified, by target pedestrian
Between influence each other and be expressed as causing relationship, promotion relationship, obstruction relationship；
Step 2, momentum kinetic model is used to construct the mathematical model in the target line human world in video, direction initialization threshold
Value, and the causality decision condition based on momentum kinetic model is proposed based on direction threshold value；
Step 3, momentum kinetic model is used to construct the mathematical model in the target line human world in video, is obtained by calculation
The cause and effect value γ in the target line human world identifies the causality in the target line human world according to the magnitude range of cause and effect value.
Further, the specific method is as follows for the step 1:
Step 1.1, each target pedestrian is tracked, the motion profile of each target pedestrian is obtained and to calculate target initial
Momentum；
It can get the location information of pedestrian α according to Cam-shift track algorithm, calculate the position of the every two interframe pedestrian of pedestrian α
Set difference, the i.e. movement velocity of each frame of pedestrian；Target pedestrian's initial momentumOccurrence can be acquired according to formula (1)
Wherein, Oj(j=i ..., n) indicates target centroid position in jth frame image, can be obtained by target tracking algorism；n
Indicate video sequence length；T indicates the time of every frame；The quality of m expression pedestrian；The initial position of i expression video search frame；j
Indicate the present frame of target pedestrian；
Wherein, Oj(j=i ..., n) indicates target centroid position in jth frame image, can be obtained by target tracking algorism；n
Indicate video sequence length, occurrence can be determines according to actual conditions；T indicates the time of every frame；M indicates the quality of pedestrian, leads to
Often removing average value is 60kg；The initial position of i expression video search frame；The present frame of j expression target pedestrian；
Step 1.2, target pedestrian is calculated by the momentum of other pedestrian's repulsive forces
Basic model using the psychological forces in social force as impulse force, formula are as follows:
Wherein, AαAnd BαIt is constant, respectively indicates the interaction strength and sphere of action of pedestrian α Yu other pedestrians；rαβ
Be interaction two pedestrians α and β radius and；dαβIt is the distance between pedestrian α and β；nαβIt is that pedestrian α is directed toward by pedestrian β
Unit vector, nαβAnd dαβIt changes over time.Model provides that the radius of each pedestrian is determined by the space that pedestrian occupies.AαTable
It is power, unit N up to mode；The radius of pedestrian takes being uniformly distributed for shoulder breadth [0.5,0.7]；Constant BαIt, can root for range constants
The actual conditions value factually tested.
In view of pedestrian is influenced direction difference by other people, increase direction weight g (λ)；
Wherein, λ is direction weight coefficient, can value according to actual needs.For pedestrian movement direction and suffered impulse force direction
Angle.
When distance is excessive between pedestrian, without influence between two pedestrians, impulse force should be 0 at this time, and momentumIt can obtain
Final impulsive function model is as follows:
In formula, l is causal influence range parameter, and value is determines according to actual conditions.For unit direction vector.Δ t is
The reaction time of one people, the reaction time of common people are 0.15~0.5s, and the perception-reaction time of people is not normal state point
Cloth, logarithm are in the normal distribution approached, and the average perception-reaction time of people is generally 0.4s.
Step 1.3, the prediction end momentum of target pedestrian is calculatedWith practical last momentum
Practical end momentumIt can be calculated by formula (1)；Based on principle of conservation of momentum, pedestrian α is analyzed pedestrian β's
The change of motion state under the influence of impulse force,
It obtains predicting last momentum by formula (5)
In formula,For target pedestrian's initial momentum；It is target pedestrian by the momentum of other pedestrian's repulsive forces.
Further, specific step is as follows for step 2:
Step 2.1, momentum kinetic model is used to construct the mathematical model in the target line human world in video；
Video sequence length is h, and length is set in entire sequence as the translation window of s, is calculated in the translation window
Four key elements being previously mentioned in momentum model, i.e.,WithFirstly, n before being chosen in translation window
Frame, for calculatingThe most intermediate frame of translation window is chosen later as present frame, is calculatedIt is obtained then according to formula (5)The last n frame for finally choosing translation window, calculates according to formula (1)
Step 2.2, direction threshold concept is proposed；
If 4 elements --- the initial momentum of target pedestrianMomentum of the target pedestrian by other pedestrian's repulsive forcesIn advance
Survey last momentumPractical end momentumThe angle between element is in direction threshold θ two-by-twotIn range, then two pedestrian movements
The vector in direction be considered as it is equidirectional, be otherwise considered as not in the same direction；
In order to judge direction, the angle between several vectors is sought first, is calculated as follows:
It can be acquired respectively by formula (6) (7) (8)WithAngle thetax；WithAngle thetaz；WithFolder
Angle θy；Enable x=θx-θt, y=θy-θt, z=θz-θt；θtFor direction threshold value；When equidirectional, x, y, z≤0, on the contrary x, y, z > 0；
Step 2.3, the causality decision condition based on momentum kinetic model is proposed based on direction threshold value.
Further, specific step is as follows for the step 3:
To x, the value of y is normalized, and keeps positive and negative values range equal；
Shown in the calculation method such as formula (11) of cause and effect value γ
The codomain of obtained γ is [- 360 °, 360 °], more intuitive for performance, is further processed to γ, by γ's
Value narrows down to [0 °, 360 °], as shown in formula (12)
Three kinds of causalities and the relationship of cause and effect value γ ' are as follows:
The codomain range for leading to cause and effect value γ ' when relationship is [315 °, 360 °]；
The codomain range [45 °, 90 °] of cause and effect value γ ' when promotion relationship；
When obstruction relationship the codomain range of cause and effect value γ ' be (225 °, 270 °]；
There is no the codomain range of cause and effect value γ ' when causality [135 °, 225 °].
Compared with prior art, the present invention has the advantage that proposing that momentum is dynamic in conjunction with momentum theorem and social force model
Mechanical model overcomes existing model and is difficult to in reality the shortcomings that the quantificational description of pedestrian's relationship to lines of description interpersonal relation,
And cause and effect value is converted by complicated causality determination method, to keep causal identification more intuitive and accurate.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the present invention.
Fig. 2 is the specific implementation flow chart of steps of the method for the present invention.
Fig. 3 is the momentum kinetic model figure of the method for the present invention.
Fig. 4 is the embodiment explanatory diagram of the method for the present invention.
Fig. 5 is the embodiment result figure of the method for the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawing:
Causality of the present invention includes leading to three kinds of relationship, promotion relationship, obstruction relationship relationships, and relationship is caused to be
Feeling the pulse with the finger-tip mark pedestrian changes its initial direction of motion due to the influence of extraneous factor；Promotion relationship refers to the effect of external force to mesh
Mark pedestrian plays the role of promotion and final and have not been changed its direction；Obstruction relationship refers to the movement of target pedestrian by the external world
The influence that factor hinders but and has not been changed its initial motion direction；Target pedestrian's initial momentumTarget pedestrian is by other pedestrians
The momentum of repulsive forcePredict last momentumLast momentum obtained by actual measurement4 elements as target pedestrian；
As shown in Figure 1 and Figure 2, the recognition methods the following steps are included:
Step 1, video image is obtained, using cam shift algorithm, obtains the position of each target pedestrian in video image
Set, speed and range information, in conjunction in social force psychological forces model and momentum theorem construct momentum kinetic model；It utilizes
Relationship between 4 elements of target pedestrian constructs identification condition, the causality in the two target line human world is identified, by target pedestrian
Between influence each other and be expressed as causing relationship, promotion relationship, obstruction relationship.
Step 1.1, each target pedestrian is tracked, the motion profile of each target pedestrian is obtained and to calculate target initial
Momentum；
It can get the location information of pedestrian α according to Cam shift track algorithm, calculate the position of the every two interframe pedestrian of pedestrian α
Set difference, the i.e. movement velocity of each frame of pedestrian；Target pedestrian's initial momentumOccurrence can be acquired according to formula (1)
Wherein, Oj(j=i ..., n) indicates target centroid position in jth frame image, can be obtained by target tracking algorism；n
Indicate video sequence length, occurrence can be determines according to actual conditions；T indicates the time of every frame；M indicates the quality of pedestrian, leads to
Often removing average value is 60kg；The initial position of i expression video search frame；The present frame of j expression target pedestrian.
Step 1.2, target pedestrian is calculated by the momentum of other pedestrian's repulsive forces
Basic model using the psychological forces in social force as impulse force, formula are as follows:
In view of pedestrian is influenced direction difference by other people, increase direction weight g (λ)；
When distance is excessive between pedestrian, without influence between two pedestrians, impulse force should be 0 at this time, and momentumIt can obtain
Final impulsive function model is as follows:
In formula, l is causal influence range parameter, and value is determines according to actual conditions.For unit direction vector.Δ t is
The reaction time of one people, the reaction time of common people are 0.15~0.5s, and the perception-reaction time of people is not normal state point
Cloth, logarithm are in the normal distribution approached, and the average perception-reaction time of people is generally 0.4s.
Step 1.3, the prediction end momentum of target pedestrian is calculatedWith last momentum obtained by actual measurement
Last momentum obtained by actual measurementIt can be calculated by formula (1)；Based on principle of conservation of momentum, analyzes pedestrian α and exist
The change of motion state under the influence of the impulse force of pedestrian β,
It obtains predicting last momentum by formula (5)
In formula,For target pedestrian's initial momentum；It is target pedestrian by the momentum of other pedestrian's repulsive forces.
Step 2, momentum kinetic model is used to construct the mathematical model in the target line human world in video, direction initialization threshold
Value, and the causality decision condition based on momentum kinetic model is proposed based on direction threshold value；
Step 2.1, momentum kinetic model is used to construct the mathematical model in the target line human world in video；
Video sequence length is h, and length is set in entire sequence as the translation window of s, is calculated in the translation window
Four key elements being previously mentioned in momentum model, i.e.,WithFirstly, n before being chosen in translation window
Frame, for calculatingThe most intermediate frame of translation window is chosen later as present frame, is calculatedIt is obtained then according to formula (5)The last n frame for finally choosing translation window, calculates according to formula (1)
Step 2.2, direction threshold concept is proposed；
The equidirectional finger direction angle of two vectors is 0 ° on stricti jurise.And according to the pedestrian movement direction of practical measurement
Being impossible proper consistent, therefore set forth herein the concepts of direction threshold value, it is believed that in direction threshold θtIn, two
Vector is considered as equidirectional.
If 4 elements --- the initial momentum of target pedestrianMomentum of the target pedestrian by other pedestrian's repulsive forcesIn advance
Survey last momentumLast momentum obtained by actual measurementThe angle between element is in direction threshold θ two-by-twotIn range, then two
The vector in pedestrian movement direction be considered as it is equidirectional, be otherwise considered as not in the same direction；
In order to judge direction, the angle between several vectors is sought first, is calculated as follows:
It can be acquired respectively by formula (6) (7) (8)WithAngle thetax；WithAngle thetaz；WithFolder
Angle θy；Enable x=θx-θt, y=θy-θt, z=θz-θt；θtFor direction threshold value；When equidirectional, x, y, z≤0, on the contrary x, y, z > 0；
Step 2.3, the causality decision condition based on momentum kinetic model is proposed based on direction threshold value, such as 1 institute of table
Show.
Table 1 is based on momentum kinetic model causality and judges table
Step 3, momentum kinetic model is used to construct the mathematical model in the target line human world in video, is obtained by calculation
The cause and effect value γ in the target line human world identifies the causality in the target line human world according to the magnitude range of cause and effect value.
According to angle thetaiThe value range of (i=x, y) is [0,180], it is known that the value range of x and y is [- θt, 180-
θt].It can be seen that at this time the value range of x and y be it is positive and negative asymmetric, positive and negative values range is according to direction threshold θtVariation and
Change.So to normalize to x, the value of y, keep positive and negative values range equal.
In order to which more intuitive expression causality identification is as a result, this paper presents the concept of cause and effect value, cause and effect value
Size can indicate causal classification and variation tendency between pedestrian.Shown in circular such as formula (11).
The codomain of obtained γ is [- 360 °, 360 °] in order to which the more intuitive of performance is herein further processed γ, by γ
Value range narrow down to [0 °, 360 °], as shown in formula (12):
Three kinds of causality codomain ranges are as shown in table 2.
Table 2
| Causality | γ’ |
| Cause | (315 °, 360 °] |
| Promote | (45 °, 90 °] |
| It hinders | (225 °, 270 °] |
| It is undefined | others |
| It is not present | [135 °, 225 °] |
Groupuscule is constructed, target number is five, as shown in Figure 4.Pedestrian α is this object of experiment, other four pedestrian's packets
Pedestrian 1, pedestrian 2, pedestrian 3, pedestrian 4 are included, four curves are respectively pedestrian 1, pedestrian 2, pedestrian 3, pedestrian 4 and target line in Fig. 5
Causality value between people α.Before 168 frames, four curve co-insides, all value is 0, since 169 frames, the value of dotted curve
Between 225~270, the causality between pedestrian 1 and object of experiment is to hinder at this time, until the 259th frame, since two pedestrians hand over
Mistake separates, and no longer there is causality between target.Pedestrian 2 is from 314 frames to 331 frames, cause and effect value cause and effect between [45,90] and target
Relationship is to cause, and from 323 frames to 360 frames, the two causality becomes promoting.Pedestrian 4 is from 335 frames to 337 frames and target pedestrian
Causality is to hinder, and the cause and effect value of pedestrian 3 is 0 always, and causality is not present with the target line human world.The result of the present embodiment
As shown in Figure 5.
Embodiment described above only describe the preferred embodiments of the invention, not to model of the invention
It encloses and is defined, without departing from the spirit of the design of the present invention, those of ordinary skill in the art are to technical side of the invention
The various changes and improvements that case is made should all be fallen into the protection scope that claims of the present invention determines.
Claims (4)
1. causality recognition methods between a kind of pedestrian based on momentum kinetic model, it is characterised in that: causality includes
Lead to three kinds of relationship, promotion relationship, obstruction relationship relationships, relationship is caused to refer to that target pedestrian changes due to the influence of extraneous factor
Its initial direction of motion；Promotion relationship refer to the effect of external force to target pedestrian play the role of promotion and it is final not
Change its direction；Obstruction relationship refers to influence that the movement of target pedestrian is hindered by extraneous factor but and has not been changed its and initial transport
Dynamic direction；Target pedestrian's initial momentumMomentum of the target pedestrian by other pedestrian's repulsive forcesPredict last momentumIt is real
Border end momentum4 elements as target pedestrian；
The recognition methods the following steps are included:
Step 1, video image is obtained, using cam-shift algorithm, obtains the position of each target pedestrian, speed in video image
Degree and range information, in conjunction with the psychological forces model and momentum theorem building momentum kinetic model in social force；Utilize target
Relationship between 4 elements of pedestrian constructs identification condition, the causality in the two target line human world is identified, by the target line human world
It influences each other and is expressed as causing relationship, promotion relationship, obstruction relationship；
Step 2, momentum kinetic model is used to construct the mathematical model in the target line human world in video, direction initialization threshold value, and
The causality decision condition based on momentum kinetic model is proposed based on direction threshold value；
Step 3, momentum kinetic model is used to construct the mathematical model in the target line human world in video, target is obtained by calculation
Cause and effect value γ ' between pedestrian identifies the causality in the target line human world according to the magnitude range of cause and effect value.
2. causality recognition methods between a kind of pedestrian based on momentum kinetic model according to claim 1, special
Sign is that the specific method is as follows for the step 1:
Step 1.1, each target pedestrian is tracked, the motion profile of each target pedestrian is obtained and calculates target just initiating
Amount；
It can get the location information of pedestrian α according to Cam-shift track algorithm, calculate the alternate position spike of the every two interframe pedestrian of pedestrian α
Value, the i.e. movement velocity of each frame of pedestrian；Target pedestrian's initial momentumOccurrence can be acquired according to formula (1)
Wherein, Oj(j=i ..., n) indicates target centroid position in jth frame image, can be obtained by target tracking algorism；N indicates view
Frequency sequence length；T indicates the time of every frame；The quality of m expression pedestrian；The initial position of i expression video search frame；J indicates mesh
Mark the present frame of pedestrian；
Step 1.2, target pedestrian is calculated by the momentum of other pedestrian's repulsive forces
Basic model using the psychological forces in social force as impulse force, formula are as follows:
Wherein AαAnd BαIt is constant, respectively indicates the interaction strength and sphere of action of pedestrian α Yu other pedestrians；rαβIt is mutual
Effect two pedestrians α and β radius and；dαβIt is the distance between pedestrian α and β；nαβIt is the unit that pedestrian α is directed toward by pedestrian β
Vector, nαβAnd dαβIt changes over time；Model provides that the radius of each pedestrian is determined by the space that pedestrian occupies；AαExpression way
For power, unit N；The radius of pedestrian takes being uniformly distributed for shoulder breadth [0.5,0.7]；Constant BαIt, can be according to experiment for range constants
Actual conditions value；
In view of pedestrian is influenced direction difference by other people, increase direction weight g (λ)；
Wherein, λ is direction weight coefficient, can value according to actual needs；For the folder in pedestrian movement direction and suffered impulse force direction
Angle；
When distance is excessive between pedestrian, without influence between two pedestrians, impulse force should be 0 at this time, and momentumIt can obtain finally
Impulsive function model is as follows:
In formula, l is causal influence range parameter；For unit direction vector；Δ t is the reaction time of people, general person's development
Time is 0.15~0.5s, and the perception-reaction time of people is not normal distribution, and logarithm is in the normal distribution approached, people's
Average perception-reaction time is generally 0.4s；
Step 1.3, the prediction end momentum of target pedestrian is calculatedWith practical last momentum
Practical end momentumIt can be calculated by formula (1)；Based on principle of conservation of momentum, pedestrian α is analyzed in the impulse force of pedestrian β
Under the influence of motion state change,
It obtains predicting last momentum by formula (5)
In formula,For target pedestrian's initial momentum；It is target pedestrian by the momentum of other pedestrian's repulsive forces.
3. causality recognition methods between a kind of pedestrian based on momentum kinetic model according to claim 1, special
Sign is that specific step is as follows for step 2:
Step 2.1, momentum kinetic model is used to construct the mathematical model in the target line human world in video；
Video sequence length is h, and length is set in entire sequence as the translation window of s, calculates momentum in the translation window
Four key elements being previously mentioned in model, i.e.,WithFirstly, n frame before being chosen in translation window,
For calculatingThe most intermediate frame of translation window is chosen later as present frame, is calculatedIt is obtained then according to formula (5)The last n frame for finally choosing translation window, calculates according to formula (1)
Step 2.2, direction threshold concept is proposed；
If 4 elements --- the initial momentum of target pedestrianMomentum of the target pedestrian by other pedestrian's repulsive forcesPrediction end
MomentumPractical end momentumThe angle between element is in direction threshold θ two-by-twotIn range, then two pedestrian movement directions
Vector be considered as it is equidirectional, be otherwise considered as not in the same direction；
In order to judge direction, the angle between several vectors is sought first, is calculated as follows:
It can be acquired respectively by formula (6) (7) (8)WithAngle thetax；WithAngle thetaz；WithAngle
θy；Enable x=θx-θt, y=θy-θt, z=θz-θt；θtFor direction threshold value；When equidirectional, x, y, z≤0, on the contrary x, y, z > 0；
Step 2.3, the causality decision condition based on momentum kinetic model is proposed based on direction threshold value.
4. causality recognition methods between a kind of pedestrian based on momentum kinetic model according to claim 3, special
Sign is that specific step is as follows for the step 3:
To x, the value of y is normalized, and keeps positive and negative values range equal；
Shown in the calculation method such as formula (11) of cause and effect value γ
γ is further processed, the value of γ is narrowed down into [0 °, 360 °], as shown in formula (12)
The codomain range for leading to cause and effect value γ ' when relationship is [315 °, 360 °]；
Three kinds of causalities and the relationship of cause and effect value γ ' are as follows:
The codomain range [45 °, 90 °] of cause and effect value γ ' when promotion relationship；
When obstruction relationship the codomain range of cause and effect value γ ' be (225 °, 270 °]；
There is no the codomain range of cause and effect value γ ' when causality [135 °, 225 °].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610973556.8A CN106529465B (en) | 2016-11-07 | 2016-11-07 | Causality recognition methods between a kind of pedestrian based on momentum kinetic model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610973556.8A CN106529465B (en) | 2016-11-07 | 2016-11-07 | Causality recognition methods between a kind of pedestrian based on momentum kinetic model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106529465A CN106529465A (en) | 2017-03-22 |
| CN106529465B true CN106529465B (en) | 2019-10-11 |
Family
ID=58349620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610973556.8A Active CN106529465B (en) | 2016-11-07 | 2016-11-07 | Causality recognition methods between a kind of pedestrian based on momentum kinetic model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106529465B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256435B (en) * | 2017-12-25 | 2019-10-11 | 西安电子科技大学 | Based on the causal video behavior recognition methods of component |
| CN108363988A (en) * | 2018-03-09 | 2018-08-03 | 燕山大学 | A kind of people counting method of combination characteristics of image and hydrodynamics characteristic |
| CN109064484B (en) * | 2018-03-21 | 2022-02-08 | 南京邮电大学 | Crowd movement behavior identification method based on fusion of subgroup component division and momentum characteristics |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020591A (en) * | 2012-11-21 | 2013-04-03 | 燕山大学 | Medium scale crowd abnormal behavior detection method based on causal network analysis |
| CN105243673A (en) * | 2015-08-05 | 2016-01-13 | 深圳大学 | Motion estimation method based on block matching, motion estimation system and application thereof |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100080298A1 (en) * | 2008-09-30 | 2010-04-01 | Hsueh-Ming Hang | Refined Weighting Function and Momentum-Directed Genetic search pattern algorithm |
-
2016
- 2016-11-07 CN CN201610973556.8A patent/CN106529465B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020591A (en) * | 2012-11-21 | 2013-04-03 | 燕山大学 | Medium scale crowd abnormal behavior detection method based on causal network analysis |
| CN105243673A (en) * | 2015-08-05 | 2016-01-13 | 深圳大学 | Motion estimation method based on block matching, motion estimation system and application thereof |
Non-Patent Citations (5)
| Title |
|---|
| A modelling-mapping approach for fine-scale assessment of pedestrian-level wind in high-density cities;Yuan, Chao 等;《BUILDING AND ENVIRONMENT》;20160229;第152-165页 * |
| Causal reasoning with forces;Wolff, Phillip 等;《FRONTIERS IN HUMAN NEUROSCIENCE》;20150131;第1-21页 * |
| Small scale crowd behavior classification by Euclidean distance variation-weighted network;Zhang, Xuguang 等;《MULTIMEDIA TOOLS AND APPLICATIONS》;20161031;第11945-11960页 * |
| 基于反馈的群体疏散行为定性预测;邵晨曦 等;《系统仿真学报》;20040430;第978-981页 * |
| 基于速度场简化的人群行为分析;李楠 等;《计算机与通信技术》;20141031;第490-496页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106529465A (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kim et al. | Simultaneous gesture segmentation and recognition based on forward spotting accumulative HMMs | |
| CN104599292B (en) | A kind of anti-noise moving object detection algorithm decomposed based on low-rank matrix | |
| CN104050488B (en) | A kind of gesture identification method of the Kalman filter model based on switching | |
| CN103389799B (en) | A kind of opponent's fingertip motions track carries out the method for following the tracks of | |
| CN106529465B (en) | Causality recognition methods between a kind of pedestrian based on momentum kinetic model | |
| CN106325485B (en) | A kind of gestures detection recognition methods and system | |
| CN110458095A (en) | A kind of recognition methods, control method, device and the electronic equipment of effective gesture | |
| CN104573621A (en) | Dynamic gesture learning and identifying method based on Chebyshev neural network | |
| CN105912985A (en) | Human skeleton joint point behavior motion expression method based on energy function | |
| CN107301376B (en) | Pedestrian detection method based on deep learning multi-layer stimulation | |
| CN101251928A (en) | Object tracking method based on core | |
| CN110298263B (en) | Real-time accurate and contactless gesture recognition method and system based on RFID system | |
| CN114118167B (en) | Action sequence segmentation method aiming at behavior recognition and based on self-supervision less sample learning | |
| CN105809713A (en) | Object tracing method based on online Fisher discrimination mechanism to enhance characteristic selection | |
| CN103198330A (en) | Real-time human face attitude estimation method based on depth video streaming | |
| CN105447506A (en) | Gesture recognition method based on interval distribution probability characteristics | |
| CN109684990A (en) | A kind of behavioral value method of making a phone call based on video | |
| Li | Design of human-computer interaction system using gesture recognition algorithm from the perspective of machine learning | |
| CN103020631B (en) | Human movement identification method based on star model | |
| CN109993188A (en) | Data label recognition methods, Activity recognition method and device | |
| CN112052786A (en) | Behavior prediction method based on grid division skeleton | |
| CN103826202A (en) | Method for improving WiFi positioning result jumping based on cell phone sensor information | |
| Sheng et al. | Context-Aware Faster RCNN for CSI-Based Human Action Perception | |
| Wang et al. | Classifying exotic wakes with a flow speed sensor | |
| Wang et al. | Gesture recognition based on BoF and its application in human-machine interaction of service robot |
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 |