CN105913080B - Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix - Google Patents
Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix Download PDFInfo
- Publication number
- CN105913080B CN105913080B CN201610216834.5A CN201610216834A CN105913080B CN 105913080 B CN105913080 B CN 105913080B CN 201610216834 A CN201610216834 A CN 201610216834A CN 105913080 B CN105913080 B CN 105913080B
- Authority
- CN
- China
- Prior art keywords
- class
- model
- state
- target
- moment
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
Abstract
The invention discloses one kind based on the motor-driven non-elliptical extension target joint tracking of random matrix and classification method, mainly solves the problems, such as that the prior art cannot handle motor-driven non-elliptical extension target joint tracking and classification.Implementation step is: firstly, non-elliptical extension target is divided into multiple oval sub-goals, and indicating its structural information with the relativeness of sub-goal；Secondly, being filtered with multi-model process to sub-goal under Bayesian frame based on the mode for describing sub-goal state with random matrix；Finally, according to the class state of the motion state of the structural information real-time estimation sub-goal between filter result and sub-goal, extended mode and non-elliptical extension target.Emulation experiment shows that the present invention efficiently solves motor-driven non-elliptical extension target joint tracking and classification problem, can be used for Target Tracking System.
Description
Technical field
The invention belongs to field of information processing, in particular to a kind of extension target joint tracking and classification method can be used for
Target Tracking System.
Background technique
In recent years, it with the continuous improvement of the sensor resolutions such as radar, infrared, extends target following technology and has caused state
The extensive concern of inside and outside researcher.The technology is not only in the military neck such as missile defence, aerial reconnaissance and early warning, battlefield surveillance
Domain, and also have broad application prospects in civil fields such as robot vision, air traffic navigation and control.Extend target
Refer to: due to being closer between the raising or target and sensor of sensor resolution, the echo-signal of single target may
It falls into multiple resolution cells, causes the different equivalent scattering center of the target that may generate multiple measurements simultaneously.According to acquisition
The difference of measurement information, extension target can be further divided into elliptical extension target and non-two class of elliptical extension target.Wherein, non-ellipse
Circle extension target is since the flexibility of model can describe the target of various irregular shapes, but corresponding joint tracks and divides
Class method also becomes considerably complicated therewith, especially maneuvering target, realizes that difficulty is bigger.
Currently, mainly having for non-elliptical extension target following with classification method: based on the non-elliptical extension mesh of random matrix
Mark tracking and based on the non-elliptical extension target joint tracking of random matrix and classification method.Wherein, first method is with shellfish
Ye Si is filtered into basic framework, and with the non-elliptical extension target of multiple oval sub-goal approximations, sub-goal is described using random matrix,
Can real-time estimation target motion state and extended mode.Meanwhile for the motor-driven problem of processing target, this method joined multimode
Type.However, due to this method be related to multi-model and model and target and measure between related question, computation complexity compared with
Height is not suitable for real-time modeling method system.In addition, for convenience's sake, this method also has ignored the estimation of target class state.
For this purpose, Lan et al. was proposed in 2014 based on the non-elliptical extension target joint tracking of random matrix and classification side
Method.This method joined the priori structural information of class unlike first method, including the space between oval sub-goal
Relationship.In addition, this method only estimates main elliptical state for simplified model.Compared to first method, this method calculates letter
Single, error is smaller, it is easier to realize.But for convenience's sake, this method have ignored non-elliptical extension target occur it is motor-driven
Situation causes the evaluated error of target state, extended mode and class state at this time larger, may not apply in this way
Based on the motor-driven non-elliptical extension target joint tracking of random matrix and classification.
Summary of the invention
It is an object of the invention in view of the above-mentioned problems, proposing a kind of based on the motor-driven non-elliptical extension target connection of random matrix
Tracking and classification method are closed, to handle the motor-driven situation of non-elliptical extension target following, reduces the evaluated error of target, at raising
The estimated accuracy of non-elliptical extension dbjective state when motor-driven.
Technical thought of the invention is: firstly, non-elliptical extension target is divided into multiple oval sub-goals, and passing through son
Relativeness between target indicates its structural information；Secondly, based on the mode for describing sub-goal state with random matrix, in shellfish
Sub-goal is filtered with multi-model process under this frame of leaf；Finally, being believed according to the structure between filter result and sub-goal
Cease the class state of the motion state of real-time estimation sub-goal, extended mode and non-elliptical extension target.Implementation step includes
It is as follows:
Based on the motor-driven non-elliptical extension target joint tracking of random matrix and classification method, comprising:
(1) dbjective state at k-1 moment is initializedInitialize j-th of model of the i-th class probability and the i-th class
Model probability is respectively as follows:WithWherein,It is the motion state of target in j-th of model of the i-th class,It is the i-th class
The extended mode of target, j=1 ..., N in j-th of model, N indicate pattern number, i=1 ..., nc, ncIt is class number, k=1；
(2) in k >=1, by nkThe measurement that a k moment is collected into is divided intoGroup, and it is seen as an entirety by each group,
Measured the correlating event number between son ellipse are as follows:WhereinFor sub oval number,
(3) to the dbjective state at k-1 momentIt is reinitialized, obtains stateIts
In,It is the target state after being reinitialized in j-th of model of the i-th class,It is weight in j-th of model of the i-th class
Target extended mode after new initialization；
(4) for the state after reinitializingIt is filtered, j-th of model of the i-th class after being filtered
More new state under correlating event lWith corresponding model likelihoodWherein,WithRespectively i-th
Target update motion state and target update extended mode under j-th of model interaction event l of class；It further calculates to obtain i-th
The model likelihood of j-th of model of classTarget update state corresponding with the k moment
Wherein,The updated target state of state in i-th j-th of class model,Shape in i-th j-th of class model
The updated target extended mode of state；
(5) according to known likelihoodWith the model probability of k-1 moment j-th of model of the i-th classCalculate the k moment i-th
The model probability of j-th of model of class
(6) by the model probability at k momentAnd dbjective stateThe dbjective state of the i-th class is calculated It is target state in the i-th class,It is target extended mode in the i-th class；
(7) according to the model likelihood of j-th of model of the i-th classWith corresponding model probabilityCalculate the similar of the i-th class
SoAnd according to the similar right of the i-th classWith the i-th class probability at -1 moment of kthThe i-th class probability at k moment is calculatedAnd it exports；
(8) according to the motion state in the i-th classAnd extended modeCalculate all elliptical movement shapes of son in the i-th class
StateAnd extended modeAnd by the two statesWith the probability of class iProbability weight calculating is carried out, son is obtained
The state estimation of oval sEstimate with extended modeAnd it exports；
(9) judge whether tracking terminates, if the target of input subsequent time measures, enable k=k+1, return step (2) is right
The dbjective state of subsequent time is estimated that otherwise, object tracking process terminates.
The invention has the following advantages that
1) present invention is approximately no longer an ellipse target due to being with non-elliptical extension dbjective state descriptive model,
But be approximately multiple ellipses, so the extended mode of energy good fit target complexity, preferably reduces losing for useful information
It loses.
2) present invention is due to being to combine tracking and classification method with non-elliptical extension target, and incorporated Interactive Multiple-Model
Mechanism, thus can preferable the solution tracking of elliptical extension target joint and target in classification by no means motor-driven problem, can not only be preferable
Ground tracks motor-driven non-elliptical extension target, and correct target classification state can be provided in target maneuver.
Detailed description of the invention
Fig. 1 is overall flow figure of the invention；
The non-ellipse target structure that Fig. 2 simplifies
Fig. 3 is target trajectory figure used in emulation；
Fig. 4 is tracking result partial enlarged view in part of the present invention；
Fig. 5 is the present invention and the position mean square error comparison diagram based on the non-elliptical extension method for tracking target of random matrix；
Fig. 6 is the present invention and the speed mean square error comparison diagram based on the non-elliptical extension method for tracking target of random matrix；
Fig. 7 is that the present invention is compared with the extended mode mean square error based on the non-elliptical extension method for tracking target of random matrix
Figure；
Fig. 8 is class probabilistic simulation result of the present invention.
Specific embodiment
Referring to Fig.1, specific implementation of the invention the following steps are included:
Model probability under the conditions of step 1. initialized target state estimation, class probability and class.
1.1) dbjective state at k-1 moment is initializedWherein,It is that target is transported in j-th of model of the i-th class
Dynamic state,Target extended mode, j=1 ..., N in j-th of model of the i-th class, N indicates Number of Models, i=1 ...,
nc, ncIt is class number, initial time k=1；
1.2) model probability for initializing j-th of model of the i-th class probability and the i-th class is respectively as follows:With
Step 2., which measures, to be divided.
In k >=1, by nkThe measurement that a k moment is collected into is divided into K-means methodGroup, whereinIt will
Each group is seen as an entirety, calculates the correlating event number measured between son ellipse are as follows:WhereinIt is ellipse for son
Circle number.
Step 3. reinitializes.
To the dbjective state at k-1 momentIt is reinitialized, the dbjective state after being reinitializedWherein,It is the target state after reinitializing,It is the target extension after reinitializing
State.
Step 4. model filtering.
4.1) dbjective state after being reinitialized to the k-1 momentA step transfer is carried out, the i-th class is obtained
J simulated target motion prediction statePredicted state is extended with corresponding target
Wherein,It is in j-th of model of the i-th classWithMapping function one by one, ψj|i() is i-th
In j-th of model of classWithMapping function one by one；
4.2) according to target prediction stateThe observation being collected into the k moment calculates the i-th class of current time the
Dbjective state in j model under correlating event lAnd corresponding model likelihoodWherein For the target update motion state under correlating event l,For the target update extended mode under correlating event l:
Wherein,Indicate n in k moment correlating event lkA measurement allocation result, φ(j|i)l() is the i-th class j-th
In model under correlating event lWithMapping function one by one, Φ(j|i)lIt is associated in j-th of model of () i-th class
Under event lWith Mapping function one by one；
4.3) estimate correlating event l drag likelihood in j-th of model of the i-th class
Wherein, f () isWithMapping function one by one；
4.4) according to the relevant event target state of k moment j-th of model of the i-th classWith corresponding mould
Type likelihoodIt is calculated, whereinObtain the target update state of target k moment j-th of model of the i-th class
Wherein,For target update motion state,For target update extended mode；
4.5) to the model likelihood of all correlating events in j-th of model of the i-th classProbability weight summation is carried out, is obtained
The model likelihood of i-th j-th of class model
Step 5. model probability updates.
The model likelihood of known i-th class, j-th of modelWith the model probability of k-1 moment j-th of model of the i-th class
Obtain the model probability of k moment j-th of model of the i-th class
Wherein, ckFor normalization factor.
Step 6. model estimation fusion.
According to the model probability of k moment j-th of model of the i-th classWith corresponding dbjective stateIt is calculated
The dbjective state of i-th class
Wherein,It is target state in the i-th class,Target extended mode in i-th class.
The probability updating of step 7. class.
7.1) according to the model likelihood of j-th of model of the i-th classWith corresponding model probabilityThe i-th class is calculated
It is similar right
7.2) according to the i-th class probability at k-1 momentIt is similar with the i-th class rightI-th class at k moment is calculated
Probability
Wherein,ncFor class number.
The estimation fusion of step 8. class.
When Lan et al. was proposed based on random matrix non-elliptical extension target joint tracking in 2014 with classification method to
Go out the estimation fusion calculation method of class, circular is as follows:
8.1) according to the motion state in the i-th classAnd extended modeCalculate all elliptical movement shapes of son in the i-th class
StateAnd extended mode
8.1.1) according to the dbjective state in the i-th classCalculate the elliptical motion state of boss in the i-th classWith
Extended mode
Wherein, λ is constant；
8.1.2) according to the elliptical motion state of boss in the i-th classAnd extended modeUsing including in the i-th class
Non- oval structure information calculates all elliptical motion states of son in the i-th classAnd extended mode
Wherein, ds,iIndicate the son ellipse s of the i-th class using boss's ellipse as the coordinate of reference point,WithIt is boss respectively
Elliptical extension stateCarry out spin matrix and diagonal matrix that singular value decomposition obtains, diagonal matrixMiddle diagonal element is
With square descending arrangement of main oval half；Diagonal matrixMiddle diagonal element is the son ellipse s and boss's ellipse of the i-th class
The ratio between semiaxis, spin matrixIt is the non-oval structure ellipse s of the i-th class relative to the elliptical direction of boss, ()TIt indicates
Transposition operation is carried out to matrix.
8.2) according to the elliptical motion state of sons all in the i-th classExtended modeWith the probability of the i-th classMeter
It calculates and obtains the state estimation of sub- ellipse sEstimate with extended mode
Wherein, E [] is operation of averaging.
8.3) state estimation of sub- ellipse s is exportedEstimate with extended mode
Step 9. judges whether tracking terminates.
If the target for inputting subsequent time measures, k=k+1 is enabled, return step 2 carries out the dbjective state of subsequent time
Estimation, otherwise, object tracking process terminates.
Effect of the invention can be further illustrated by following emulation experiment:
1. simulated conditions.
Simulated environment: computer uses Intel Core i5-2400 CPU 3.1Ghz, 4GB memory, and software uses
Matlab R2012a Simulation Experimental Platform.
Emulation mode:
Method one: the method for the present invention MNEOT-JTC；
Method two: existing to be based on the motor-driven non-elliptical extension method for tracking target MNEOT of random matrix.
Simulation parameter: in simulating scenes, the non-ellipse target structure simplified is as shown in Figure 2.
Assuming that the non-elliptical extension target of known two classes, i.e. the 1st class and the 2nd class, their structure information parameter in databaseIt is respectively as follows:
1st class formation information parameter is as follows:
d1,1=[0,0]T,d2,1=[0,16.1]Tm,d3,1=-d2,1,
2nd class formation information parameter is as follows:
d1,2=[0,0]T,d2,2=[0,7.05]Tm,d3,2=-d2,2,
Wherein, spin matrix
The structure information parameter of the non-elliptical extension target of simulating scenes is set as the 1st class formation information parameter, if mesh
It marks from position [x, y]=[0,104m]TWithSpeed it is mobile, wherein x and y respectively indicates target in x
Axis and y-axis coordinate,WithRespectively indicate speed of the target in x-axis direction and y-axis direction.Sampling time interval T=0.3s.
The Poisson distribution for generating number obedience parameter beta=50 of point is measured, the position Gaussian distributed for generating point is measured.
2. emulation content and interpretation of result
Emulation experiment 1 tracks target trajectory shown in Fig. 3 in different moments with the present invention, as a result as schemed
4, in which: Fig. 4 (a) is the present invention in the partial enlarged view that motor-driven preceding moment tracking occurs；
Fig. 4 (b) is the present invention in the partial enlarged view that motor-driven moment tracking occurs；
Fig. 4 (c) is the present invention in the partial enlarged view that motor-driven rear moment tracking occurs.
Non- elliptical extension target is occurring the motor-driven preceding moment, the motor-driven moment occurs and the motor-driven rear moment occurs as seen from Figure 4
It can illustrate that the method for the present invention can preferably track motor-driven non-oval expansion in different moments preferably by multiple sub- ellipse fittings
Open up target.
Emulation experiment 2, with the present invention with the existing non-elliptical extension method for tracking target of random matrix that is based on to shown in Fig. 3
Target trajectory carries out location estimation, as a result such as Fig. 5.
From figure 5 it can be seen that the method for the present invention than it is existing had based on the non-elliptical extension method for tracking target of random matrix it is smaller
Position mean square error RMSE, illustrating that the method for the present invention is more existing is had based on the non-elliptical extension method for tracking target of random matrix
Better position estimation.
Emulation experiment 3, with the present invention with the existing non-elliptical extension method for tracking target of random matrix that is based on to shown in Fig. 3
Target trajectory carries out velocity estimation, as a result such as Fig. 6.
As seen from Figure 6, the method for the present invention than it is existing had based on the non-elliptical extension method for tracking target of random matrix it is smaller
Speed mean square error, illustrate the method for the present invention it is more existing had based on the non-elliptical extension method for tracking target of random matrix it is more preferable
Velocity estimation.
Emulation experiment 4, with the present invention with the existing non-elliptical extension method for tracking target of random matrix that is based on to shown in Fig. 3
Target trajectory is extended state estimation, as a result such as Fig. 7.
From Fig. 7 can, see the method for the present invention than it is existing had based on the non-elliptical extension method for tracking target of random matrix it is smaller
Extended mode mean square error, illustrating that the method for the present invention is more existing is had based on the non-elliptical extension method for tracking target of random matrix
Better extended mode estimation.
Emulation experiment 5 is carrying out class probability Estimation to target trajectory shown in Fig. 3 with the present invention, as a result such as Fig. 8.
As seen from Figure 8, the class probability that the class probability of the 1st class gradually tends to the 1, the 2nd class gradually tends to 0, illustrates emulation experiment
In non-elliptical extension target be divided into the 1st class, structure information parameter setting coincide in classification results and simulating scenes, shows this hair
Bright method is correct to motor-driven non-elliptical extension target classification result.
In conclusion the method for the present invention can not only preferably track motor-driven non-elliptical extension target in different moments, and
The more existing non-elliptical extension method for tracking target of random matrix that is based on of the method for the present invention is in the estimation of position, speed and extended mode
There is better performance, furthermore the method for the present invention simultaneously can also give non-elliptical extension target tracking motor-driven non-elliptical extension target
Correctly classification out.The simulation experiment result illustrates that the method for the present invention is efficiently solved based on the motor-driven non-oval expansion of random matrix
Open up the joint tracking and classification problem of target.
Claims (7)
1. joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix, comprising:
(1) dbjective state at k-1 moment is initializedInitialize the model of j-th of model of the i-th class probability and the i-th class
Probability is respectively as follows:WithWherein,It is the motion state of target in j-th of model of the i-th class,It is the i-th class j-th
The extended mode of target, j=1 ..., N in model, N indicate pattern number, i=1 ..., nc, ncIt is class number, k=1；
(2) in k >=1, by nkThe measurement that a k moment is collected into is divided intoGroup, and it is seen as an entirety by each group, it obtains
Measure the correlating event number between son ellipse are as follows:WhereinFor sub oval number,
(3) to the dbjective state at k-1 momentIt is reinitialized, obtains stateWherein,
It is the target state after being reinitialized in j-th of model of the i-th class,It is to be reinitialized in j-th of model of the i-th class
Target extended mode afterwards；
(4) for the state after reinitializingIt is filtered, j-th of model interaction thing of the i-th class after being filtered
More new state under part lWith corresponding model likelihoodWherein,WithJ-th of respectively the i-th class
Target update motion state and target update extended mode under model interaction event l；It further calculates to obtain the i-th class j-th
The model likelihood of modelTarget update state corresponding with the k moment
Wherein,The updated target state of state in i-th j-th of class model,State is more in i-th j-th of class model
Target extended mode after new；
(5) according to known likelihoodWith the model probability of k-1 moment j-th of model of the i-th classCalculate the i-th class of k moment jth
The model probability of a model
(6) by the model probability at k momentAnd dbjective stateThe dbjective state of the i-th class is calculated
It is target state in the i-th class,It is target extended mode in the i-th class；
(7) according to the model likelihood of j-th of model of the i-th classWith corresponding model probabilityCalculate the similar right of the i-th classAnd according to the similar right of the i-th classWith the i-th class probability at -1 moment of kthThe i-th class probability at k moment is calculated
And it exports；
(8) according to the motion state in the i-th classAnd extended modeCalculate all elliptical motion states of son in the i-th class
And extended modeAnd by the two statesWith the probability of class iProbability weight calculating is carried out, obtains sub- ellipse s's
State estimationEstimate with extended modeAnd it exports；
(9) judge whether tracking terminates, if the target of input subsequent time measures, enable k=k+1, return step (2) is to next
The dbjective state at moment is estimated that otherwise, object tracking process terminates.
2. according to the method described in claim 1, wherein in step (4) to reinitializing rear dbjective stateInto
Row filtering, carries out as follows:
(4a) is to j-th of simulated target state of the i-th class of k-1 momentA step transfer is carried out, j-th of mould of the i-th class is obtained
Type target prediction state
Wherein,WithIt is j-th of simulated target motion prediction state of the i-th class and target extension predicted state respectively；It is in j-th of model of the i-th classWithMapping function one by one, ψji() is in j-th of model of the i-th classWithMapping function one by one；
(4b) is according to j-th of simulated target predicted state of the i-th classIt is measured with the target that the k moment is collected into, obtains k
Target update state under moment correlating event lAnd corresponding model likelihood
Wherein,Indicate n in k moment correlating event lkA measurement allocation result, φ(ji)l() is in j-th of model of the i-th class
Under correlating event lWithMapping function one by one, Φ(ji)lCorrelating event l in j-th of model of () i-th class
UnderWithMapping function one by one, f () isWithMapping function one by one；WithTarget update motion state and target update extended mode under j-th of model interaction event l of respectively the i-th class,I-th
Correlating event l drag likelihood in j-th of model of class；
(4c) is to the target update state obtained under all correlating eventsWith corresponding model likelihoodIt is counted
It calculates, obtains the target update state of k moment j-th of model of the i-th class
Wherein,For the target update motion state of j-th of model of the i-th class,Expand for the target update of j-th of model of the i-th class
Exhibition state；
(4d) is according to the model likelihood of j-th of model of the i-th class under correlating event lWith correlating event numberObtain the i-th class
The model likelihood of j model
3. according to the method described in claim 1, wherein obtaining the model probability of k moment j-th of model of the i-th class in step (5)It is determined by following formula:
Wherein, ckFor normaliztion constant.
4. according to the method described in claim 1, wherein calculating the similar right of the i-th class of k moment in step (7)By as follows
Formula determines:
Wherein,It is the model likelihood of j-th of model of the i-th class,It is the model probability of j-th of model of the i-th class.
5. according to the method described in claim 1, the dbjective state of the i-th class is wherein calculated in step (6)Pass through
Following formula determines:
Wherein,It is the model probability of k moment j-th of model of the i-th class,The updated mesh of state in i-th j-th of class model
Motion state is marked,The updated target extended mode of state in i-th j-th of class model,It is that target moves shape in the i-th class
State,Target extended mode in i-th class.
6. according to the method described in claim 1, wherein calculating the i-th class of k moment probability in step (7)Pass through following formula
It determines:
Wherein,ncFor class number,It is the probability of i-th class at k-1 moment,It is the similar of the i-th class of k moment
So.
7. according to the method described in claim 1, wherein obtaining all elliptical movement shapes of son in the i-th class in the step (8)
StateAnd extended modeAnd calculate the state estimation of sub- ellipse sEstimate with extended modeAs follows
It determines；
(7a) is according to the motion state in the i-th classAnd extended modeCalculate the elliptical motion state of boss in the i-th classWith
Extended mode
Wherein, λ is constant；
(7b) is according to the elliptical motion state of boss in the i-th classAnd extended modeUtilize the non-ellipse for including in the i-th class
Structural information calculates all elliptical motion states of son in the i-th classAnd extended mode
Wherein, ds,iIndicate the son ellipse s of the i-th class using boss's ellipse as the coordinate of reference point,WithIt is that boss's ellipse expands respectively
Exhibition stateCarry out spin matrix and diagonal matrix that singular value decomposition obtains, diagonal matrixMiddle diagonal element is ellipse with master
Square descending arrangement of circle semiaxis；Diagonal matrixMiddle diagonal element be the i-th class son ellipse s and boss's oval half it
Than spin matrixIt is the non-oval structure ellipse s of the i-th class relative to the elliptical direction of boss, ()TIt indicates to matrix
Carry out transposition operation；
(7c) is according to the elliptical motion state of sons all in the i-th classExtended modeWith the probability of the i-th classIt calculates
To the state estimation of sub- ellipse sEstimate with extended mode:
Wherein, E [] is operation of averaging.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610216834.5A CN105913080B (en) | 2016-04-08 | 2016-04-08 | Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610216834.5A CN105913080B (en) | 2016-04-08 | 2016-04-08 | Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105913080A CN105913080A (en) | 2016-08-31 |
| CN105913080B true CN105913080B (en) | 2019-02-22 |
Family
ID=56745819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610216834.5A Active CN105913080B (en) | 2016-04-08 | 2016-04-08 | Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105913080B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106443618B (en) * | 2016-09-08 | 2019-01-04 | 北京理工大学 | A kind of high resolution radar Data Association based on standard deviation ellipse parameter auxiliary |
| CN106443661B (en) * | 2016-09-08 | 2019-07-19 | 河南科技大学 | Motor-driven extension method for tracking target based on Unscented kalman filtering |
| CN106570330A (en) * | 2016-11-08 | 2017-04-19 | 河南科技大学 | Shape estimated performance evaluation method for extended target tracing |
| US10962331B2 (en) * | 2019-06-06 | 2021-03-30 | Bae Systems Information And Electronic Systems Integration Inc. | Dynamic weapon to target assignment using a control based methodology |
| CN116736286B (en) * | 2023-05-24 | 2024-02-06 | 兰州理工大学 | Progressive Bayes extended target tracking method and system based on random hypersurface |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268567A (en) * | 2014-09-18 | 2015-01-07 | 中国民航大学 | Extended target tracking method using observation data clustering and dividing |
| KR20150026624A (en) * | 2013-09-03 | 2015-03-11 | 한국전자통신연구원 | Method for tracking location based on dynamic cluster and scheduling apparatus using the method |
| CN104730511A (en) * | 2015-04-10 | 2015-06-24 | 西安电子科技大学 | Tracking method for multiple potential probability hypothesis density expansion targets under star convex model |
| CN104794735A (en) * | 2015-04-02 | 2015-07-22 | 西安电子科技大学 | Extended target tracking method based on variational Bayesian expectation maximization |
-
2016
- 2016-04-08 CN CN201610216834.5A patent/CN105913080B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20150026624A (en) * | 2013-09-03 | 2015-03-11 | 한국전자통신연구원 | Method for tracking location based on dynamic cluster and scheduling apparatus using the method |
| CN104268567A (en) * | 2014-09-18 | 2015-01-07 | 中国民航大学 | Extended target tracking method using observation data clustering and dividing |
| CN104794735A (en) * | 2015-04-02 | 2015-07-22 | 西安电子科技大学 | Extended target tracking method based on variational Bayesian expectation maximization |
| CN104730511A (en) * | 2015-04-10 | 2015-06-24 | 西安电子科技大学 | Tracking method for multiple potential probability hypothesis density expansion targets under star convex model |
Non-Patent Citations (1)
| Title |
|---|
| 《Tracking of Maneuvering Non-Ellipsoidal Extended Object or Target Group Using Random Matrix》;Lan Jian et al;《IEEE Transactions on Signal Processing》;20140301;第62卷(第2014年第9期);全文 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105913080A (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105913080B (en) | Joint tracking and classification method based on the motor-driven non-elliptical extension target of random matrix | |
| Chen et al. | Cyber-physical system enabled nearby traffic flow modelling for autonomous vehicles | |
| CN107169435B (en) | Convolutional neural network human body action classification method based on radar simulation image | |
| CN108536171B (en) | Path planning method for collaborative tracking of multiple unmanned aerial vehicles under multiple constraints | |
| CN109633590B (en) | Extended target tracking method based on GP-VSMM-JPDA | |
| CN107193012A (en) | Intelligent vehicle laser radar multiple-moving target tracking method based on IMM MHT algorithms | |
| CN101221238B (en) | Dynamic deviation estimation method based on gauss average value mobile registration | |
| CN107066806B (en) | Data Association and device | |
| CN105699964A (en) | Road multi-target tracking method based on automobile anti-collision radar | |
| CN103678949B (en) | Density based is analyzed and many Extended target of spectral clustering follow the tracks of measurement collection division methods | |
| CN110018691A (en) | Small-sized multi-rotor unmanned aerial vehicle state of flight estimating system and method | |
| CN110825113A (en) | Formation keeping method suitable for quad-rotor unmanned aerial vehicle cluster flight | |
| CN111508282B (en) | Low-altitude unmanned farmland operation flight obstacle conflict detection method | |
| CN104793201A (en) | Modified variable-structure grid interaction multi-model filtering method for tracking hypersonic-speed target of near space | |
| CN104331623A (en) | Self-adaptive target tracking information filtering algorithm of maneuvering strategies | |
| WO2019047455A1 (en) | Multi-maneuver target tracking method and system applicable to flicker noise | |
| CN114299607A (en) | Human-vehicle collision risk degree analysis method based on automatic driving of vehicle | |
| CN107797106A (en) | A kind of PHD multiple target tracking smooth filtering methods of the unknown clutter estimations of acceleration EM | |
| Mettler | Extracting micro air vehicles aerodynamic forces and coefficients in free flight using visual motion tracking techniques | |
| Xia et al. | Extended object tracking using hierarchical truncation measurement model with automotive radar | |
| Xie et al. | Long and short term maneuver trajectory prediction of UCAV based on deep learning | |
| Lyu et al. | A model-aided optical flow/inertial sensor fusion method for a quadrotor | |
| CN113030940B (en) | Multi-star convex type extended target tracking method under turning maneuver | |
| CN109903367B (en) | Method, apparatus and computer readable storage medium for constructing map | |
| CN114462293A (en) | Method for predicting medium and long-term trajectory of hypersonic target |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |