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
- state
- model
- class
- target
- ith
- 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
- 238000000034 method Methods 0.000 title claims abstract description 69
- 239000011159 matrix material Substances 0.000 title claims abstract description 42
- 230000008569 process Effects 0.000 claims abstract description 5
- 230000000875 corresponding effect Effects 0.000 claims description 14
- 238000005259 measurement Methods 0.000 claims description 14
- 238000013507 mapping Methods 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 239000004576 sand Substances 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 2
- 241000372285 Isanda Species 0.000 claims description 2
- 230000002596 correlated effect Effects 0.000 claims description 2
- 238000000354 decomposition reaction Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 238000012546 transfer Methods 0.000 claims description 2
- 238000004088 simulation Methods 0.000 description 18
- 230000006870 function Effects 0.000 description 5
- 238000013459 approach Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Feedback Control In General (AREA)
- Image Analysis (AREA)
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 the field of information processing, and particularly relates to an extended target joint tracking and classifying method which can be used for a target tracking system.
Background
In recent years, with the increasing resolution of sensors such as radar and infrared, the expanded target tracking technology has attracted extensive attention of researchers at home and abroad. The technology has wide application prospect not only in the military fields of missile defense, aerial reconnaissance and early warning, battlefield monitoring and the like, but also in the civil fields of robot vision, aerial traffic navigation, control and the like. The extension target means: due to the increased resolution of the sensor or the closer distance between the target and the sensor, the echo signal of a single target may fall into multiple resolution cells, resulting in multiple measurements of different equivalent scattering centers of the target that may be generated simultaneously. The extended targets can be further classified into elliptical extended targets and non-elliptical extended targets according to different measurement information. The non-elliptical extended targets can describe various targets with irregular shapes due to the flexibility of the model, but the corresponding joint tracking and classification method is quite complicated, and particularly the maneuvering targets are difficult to realize.
At present, the tracking and classifying method for the non-elliptical extended target mainly comprises the following steps: a non-elliptical extended target tracking method based on a random matrix and a non-elliptical extended target joint tracking and classifying method based on the random matrix. The first method adopts Bayesian filtering as a basic frame, adopts a plurality of elliptical sub-targets to approximate a non-elliptical expansion target, adopts random matrix description for the sub-targets, and can estimate the motion state and the expansion state of the target in real time. Meanwhile, in order to deal with the target maneuver problem, the method adds multiple models. However, since this method involves multiple models and the correlation problem between the models and the targets and measurements, the computational complexity is high and it is not suitable for real-time target tracking systems. In addition, the method also ignores the estimation of the target class state for convenience.
Therefore, Lan et al proposed a non-elliptical extended target joint tracking and classification method based on a random matrix in 2014. This method differs from the first method in that a priori structural information of the class is added, including the spatial relationship between the elliptical sub-objectives. In addition, to simplify the model, the method only estimates the state of the dominant ellipse. Compared with the first method, the method is simple in calculation, small in error and easy to implement. However, for convenience, the method ignores the situation that the non-elliptical expansion target is maneuvered, so that the estimation errors of the motion state, the expansion state and the class state of the target at the moment are large, and therefore the method cannot be applied to the maneuvering non-elliptical expansion target joint tracking and classification based on the random matrix.
Disclosure of Invention
The invention aims to provide a random matrix-based maneuvering non-elliptical expansion target joint tracking and classifying method to process maneuvering conditions of non-elliptical expansion target tracking, reduce estimation errors of targets and improve estimation accuracy of non-elliptical expansion target states during maneuvering.
The technical idea of the invention is as follows: firstly, dividing a non-elliptical expansion target into a plurality of elliptical sub-targets, and expressing the structural information of the elliptical sub-targets through the relative relationship among the sub-targets; secondly, filtering the sub-target by using a multi-model method under a Bayesian framework based on a mode of describing the states of the sub-targets by using a random matrix; and finally, estimating the motion state and the expansion state of the sub targets and the class state of the non-elliptical expansion target in real time according to the filtering result and the structural information between the sub targets. The method comprises the following implementation steps:
a maneuvering non-elliptical expansion target joint tracking and classifying method based on a random matrix comprises the following steps:
(1) initializing a target state at time k-1The model probabilities of initializing the ith class probability and the ith class jth model are respectively as follows:andwherein,is the motion state of the object in the ith class jth model,is the expansion state of the target in the ith model, j is 1, N represents the number of models, i is 1c，ncIs a class number, k is 1;
(2) when k is more than or equal to 1, n is addedkThe measurement collected at each time k is divided intoAnd regarding each group as a whole, and obtaining the number of the correlation events between the measurement and the sub-ellipse as follows:whereinThe number of the sub-ellipses is,
(3) for target state at time k-1Reinitializing to obtain the stateWherein,is the target motion state after reinitialization in the jth model of the ith class,is the target expansion state after reinitialization in the ith type jth model;
(4) for the reinitialized stateFiltering to obtain the updated state of the ith type jth model correlation event l after filteringAnd corresponding model likelihoodWherein,andrespectively updating a motion state and an extended state for a target under the ith type jth model association event l; further calculating to obtain the ith classModel likelihood of jth modelTarget update status corresponding to time k
Wherein,the motion state of the target after state updating in the ith model and the jth model,updating the state of the target extension state in the ith model and the jth model;
(5) according to known likelihoodAnd model probability of ith class jth model at time k-1Calculating model probability of ith class jth model at k moment
(6) From time kModel probabilityAnd target stateCalculating to obtain the target state of the ith class Is the moving state of the object in the ith class,is the target extension state in the ith class;
(7) model likelihood based on class i jth modelAnd corresponding model probabilitiesComputing similarities to class iAnd according to the analogy of class iAnd class i probability at time k-1Calculating to obtain the i-th class probability of the k timeAnd outputting;
(8) according to motion state in class iAnd expandingExhibition stateCalculating the motion state of all sub-ellipses in the ith classAnd extended stateAnd put the two statesProbability with class iPerforming probability weighted calculation to obtain motion state estimation of sub-ellipse sAnd extended state estimationAnd outputting;
(9) and (3) judging whether the tracking is finished or not, if target measurement at the next moment is input, enabling k to be k +1, returning to the step (2) to estimate the target state at the next moment, and otherwise, finishing the target tracking process.
The invention has the following advantages:
1) the invention uses the non-ellipse to expand the object state description model, and the object is not approximated to be an ellipse, but approximated to be a plurality of ellipses, so the complex expansion state of the object can be better fitted, and the loss of useful information is better reduced.
2) The invention can better solve the maneuvering problem of the targets in the non-elliptical extended target joint tracking and classification because of the non-elliptical extended target joint tracking and classification method and the integration of an interactive multi-model mechanism, not only can better track the maneuvering non-elliptical extended target, but also can provide a correct target classification state under the maneuvering condition of the target.
Drawings
FIG. 1 is an overall flow diagram of the present invention;
FIG. 2 simplified non-elliptical target structure
FIG. 3 is a diagram of the target motion trajectory used in the simulation;
FIG. 4 is a partial enlarged view of a portion of the tracking results of the present invention;
FIG. 5 is a plot of the present invention versus the mean square error of the position of a non-elliptical extended target tracking method based on a random matrix;
FIG. 6 is a velocity mean square error comparison plot of the present invention versus a random matrix based non-elliptical extended target tracking method;
FIG. 7 is a plot of extended state mean square error versus a random matrix based non-elliptical extended target tracking method of the present invention;
FIG. 8 is the class probability simulation results of the present invention.
Detailed Description
Referring to fig. 1, a specific implementation of the present invention includes the following steps:
step 1, initializing target state estimation, class probability and model probability under class conditions.
1.1) initializing the target state at time k-1Wherein,is the motion state of the target in the ith model and the jth model,is the target expansion state in the ith model, j is 1c，ncIs a class number, and the initial time k is 1;
1.2) initializing the model probabilities of the ith class probability and the ith class jth model as follows:and
and 2, measuring and dividing.
When k is more than or equal to 1, n is addedkThe measurements collected at each time K are divided into K-meansGroup of whichConsidering each group as a whole, the number of correlation events between the measurements and the sub-ellipses is calculated as:whereinIs the number of sub-ellipses.
And 3, reinitializing.
For target state at time k-1Reinitializing to obtain reinitialized target stateWherein,is the target motion state after re-initialization,is the target extended state after re-initialization.
And 4, filtering the model.
4.1) target State after reinitialization at time k-1Performing one-step transfer to obtain the ith type jth model target motion prediction stateAnd corresponding target extended prediction states
Wherein,is in the ith class of jth modelAnda one-to-one mapping function ofj|i(. is) in the ith class of jth modelsAnda one-to-one mapping function of;
4.2) predicting the state according to the targetAnd the observation collected at the moment k is used for calculating the target state under the associated event l in the ith class jth model at the current momentAnd corresponding model likelihoodWherein The motion state is updated for the target at the associated event l,updating the extended state for the target under the associated event l:
wherein,representing n in time k associated events lkMeasured dispensing result,. phi(j|i)l(. v) is associated with event l in the jth model of class iAnda one-to-one mapping function of phi(j|i)l(. v) class i model under a correlation event /)And a one-to-one mapping function of;
4.3) estimating model likelihood under the correlated event l in the ith class and jth model
Wherein f (·,. cndot.) isAnda one-to-one mapping function of;
4.4) all the associated event target states of the ith type jth model at the moment kAnd corresponding model likelihoodPerforming a calculation in whichObtaining the target update state of the ith class jth model at the target k moment
Wherein,the motion state is updated for the target,updating the extended state for the target;
4.5) model likelihood for all associated events in class i jth modelCarrying out probability weighted summation to obtain the model likelihood of the ith model and the jth model
And 5, updating the model probability.
Model likelihood of known ith class jth modelAnd model probability of ith class jth model at time k-1Obtaining the model probability of the ith class jth model at the moment k
Wherein, ckIs a normalization factor.
And 6, model estimation fusion.
Model probability of ith class jth model according to k timeAnd corresponding target statesCalculating to obtain the target state of the ith class
Wherein,is the moving state of the object in the ith class,target extension state in class i.
And 7, updating the probability of the class.
7.1) model likelihood based on class i jth modelAnd corresponding model probabilityCalculate the similarity of the ith class
7.2) class i probability according to time k-1Similar to that of class iCalculating to obtain the i-th class probability of the k time
Wherein,ncare class numbers.
And 8, estimating and fusing the classes.
Lan et al provided a class estimation fusion calculation method when a non-elliptical extended target joint tracking and classification method based on a random matrix was proposed in 2014, and the specific calculation method is as follows:
8.1) according to the state of motion in class iAnd extended stateComputing all sub-ellipses in class iState of motion ofAnd extended state
8.1.1) based on target states in class iCalculating the motion state of the main ellipse and the sub ellipse in the ith classAnd extended state
Wherein λ is a constant;
8.1.2) according to the state of motion of the main and sub-ellipses in class iAnd extended stateCalculating the motion state of all the sub-ellipses in the ith class by using the non-ellipse structure information contained in the ith classAnd extended state
Wherein d iss,iThe coordinates of the sub-ellipse s representing the ith class with the main sub-ellipse as a reference point,andrespectively, major and minor ellipse expansion statesA rotation matrix and a diagonal matrix obtained by singular value decomposition, the diagonal matrixThe middle diagonal elements are arranged in descending order of the square of the semi-axis of the main ellipse; diagonal matrixThe middle diagonal element is the ratio of the sub-ellipse s of the ith class to the semi-axis of the main sub-ellipse, the rotation matrixIs the direction of the sub-ellipse s of the non-elliptical structure of the i-th class relative to the main sub-ellipse (·)TRepresenting a transpose operation on the matrix.
8.2) according to the motion state of all the sub-ellipses in the ith classExtended stateAnd probability of class iCalculating to obtain motion state estimation of sub-ellipse sAnd extended state estimation
Where E [. cndot. ] is the averaging operation.
8.3) estimation of the motion state of the output sub-ellipse sAnd extended state estimation
And 9, judging whether the tracking is finished or not.
And if the target measurement at the next moment is input, making k equal to k +1, returning to the step 2 to estimate the target state at the next moment, and otherwise, ending the target tracking process.
The effect of the invention can be further illustrated by the following simulation experiment:
1. and (5) simulating conditions.
Simulation environment: the computer adopts Intel Core i5-2400 CPU 3.1Ghz, 4GB memory, and the software adopts Matlab R2012a simulation experiment platform.
The simulation method comprises the following steps:
the method comprises the following steps: the method of the invention MNEOT-JTC;
the second method comprises the following steps: the conventional maneuvering non-elliptical target expansion tracking method based on a random matrix MNEOT.
Simulation parameters: in the simulation scenario, a simplified non-elliptical target structure is shown in FIG. 2.
In a hypothesis databaseTwo types of non-elliptical expansion targets are known, namely class 1 and class 2, whose structural information parametersRespectively as follows:
the class 1 structure information parameters are as follows:
d1,1＝[0,0]T,d2,1＝[0,16.1]Tm,d3,1＝-d2,1,
the class 2 structure information parameters are as follows:
d1,2＝[0,0]T,d2,2＝[0,7.05]Tm,d3,2＝-d2,2,
wherein the rotation matrix
Setting the structure information parameter of the non-elliptic extension target of the simulation scene as the type 1 structure information parameter, and setting the slave bit of the targetIs put in [ x, y ]]＝[0,104m]TTo be provided withWherein x and y represent the x-axis and y-axis coordinates of the target, respectively,andthe sampling time interval T is 0.3s, the number of measurement generation points follows a poisson distribution with parameter β being 50, and the positions of the measurement generation points follow a gaussian distribution.
2. Simulation content and result analysis
Simulation experiment 1, the target motion trajectory shown in fig. 3 is tracked at different times by using the method, and the result is shown in fig. 4, wherein: FIG. 4(a) is an enlarged partial view of the invention tracked at a point in time prior to the occurrence of a maneuver;
FIG. 4(b) is an enlarged partial view of the invention tracking at the moment of maneuver;
FIG. 4(c) is an enlarged partial view of the instant tracking of the present invention after the maneuver has occurred.
It can be seen from fig. 4 that the non-elliptical extended target can be better fitted by a plurality of sub-ellipses at the moment before the occurrence of maneuver, the moment when the occurrence of maneuver and the moment after the occurrence of maneuver, which shows that the method of the present invention can better track the maneuver non-elliptical extended target at different moments.
In simulation experiment 2, the present invention and the existing non-elliptical extended target tracking method based on random matrix are used to perform position estimation on the target motion trajectory shown in fig. 3, and the result is shown in fig. 5.
As can be seen from FIG. 5, the method of the present invention has a smaller position mean square error RMSE than the existing non-elliptical extended target tracking method based on the random matrix, which shows that the method of the present invention has a better position estimation than the existing non-elliptical extended target tracking method based on the random matrix.
Simulation experiment 3, the velocity estimation is performed on the target motion trajectory shown in fig. 3 by using the method of the present invention and the existing non-elliptical extended target tracking method based on random matrix, and the result is shown in fig. 6.
As can be seen from FIG. 6, the method of the present invention has a smaller velocity mean square error than the existing non-elliptical extended target tracking method based on the random matrix, which shows that the method of the present invention has a better velocity estimation than the existing non-elliptical extended target tracking method based on the random matrix.
Simulation experiment 4, the invention and the existing non-elliptical extended target tracking method based on random matrix are used for estimating the extended state of the target motion trail shown in fig. 3, and the result is shown in fig. 7.
From fig. 7, it can be seen that the method of the present invention has a smaller mean square error of the extended state than the existing non-elliptical extended target tracking method based on the random matrix, which shows that the method of the present invention has a better extended state estimation than the existing non-elliptical extended target tracking method based on the random matrix.
Simulation experiment 5, the result of performing class probability estimation on the target motion trajectory shown in fig. 3 by using the method of the present invention is shown in fig. 8.
As can be seen from fig. 8, the class probability of the 1 st class gradually approaches 1, and the class probability of the 2 nd class gradually approaches 0, which indicates that the non-elliptic expansion target in the simulation experiment is classified into the 1 st class, and the classification result matches with the structural information parameter setting in the simulation scene, indicating that the method of the present invention is correct for the classification result of the maneuvering non-elliptic expansion target.
In conclusion, the method not only can track the maneuvering non-elliptical expansion target better at different moments, but also has better performance in position, speed and expansion state estimation compared with the traditional random matrix-based non-elliptical expansion target tracking method. Simulation experiment results show that the method effectively solves the problem of joint tracking and classification of maneuvering non-elliptical extended targets based on the random matrix.
Claims (7)
1. A joint tracking and classifying method for maneuvering non-elliptical extended targets based on a random matrix comprises the following steps:
(1) initializing a target state at time k-1The model probabilities of initializing the ith class probability and the ith class jth model are respectively as follows:andwherein,is the motion state of the object in the ith class jth model,is the expansion state of the target in the ith model, j is 1, N represents the number of models, i is 1c，ncIs a class number, k is 1;
(2) when k is more than or equal to 1, n is addedkThe measurement collected at each time k is divided intoAnd regarding each group as a whole, and obtaining the number of the correlation events between the measurement and the sub-ellipse as follows:whereinThe number of the sub-ellipses is,
(3) for target state at time k-1Reinitializing to obtain the stateWherein,is the target motion state after reinitialization in the jth model of the ith class,is the target expansion state after reinitialization in the ith type jth model;
(4) for the reinitialized stateFiltering to obtain the updated state of the ith type jth model correlation event l after filteringAnd corresponding model likelihoodWherein,andrespectively updating a motion state and an extended state for a target under the ith type jth model association event l; further calculating to obtain the model likelihood of the ith class jth modelTarget update status corresponding to time k
Wherein,the motion state of the target after state updating in the ith model and the jth model,updating the state of the target extension state in the ith model and the jth model;
(5) according to known likelihoodAnd model probability of ith class jth model at time k-1Calculating model probability of ith class jth model at k moment
(6) From the model probability at time kAnd target stateCalculating to obtain the target state of the ith class Is the moving state of the object in the ith class,is the target extension state in the ith class;
(7) model likelihood based on class i jth modelAnd corresponding model probabilitiesComputing similarities to class iAnd according to the analogy of class iAnd class i probability at time k-1Calculating to obtain the i-th class probability of the k timeAnd outputting;
(8) according to motion state in class iAnd extended stateCalculating the motion state of all sub-ellipses in the ith classAnd extended stateAnd put the two statesProbability with class iPerforming probability weighted calculation to obtain motion state estimation of sub-ellipse sAnd extended state estimationAnd outputting;
(9) and (3) judging whether the tracking is finished or not, if target measurement at the next moment is input, enabling k to be k +1, returning to the step (2) to estimate the target state at the next moment, and otherwise, finishing the target tracking process.
2. The method of claim 1, wherein the target state after reinitialization in step (4)Filtering is carried out according to the following steps:
(4a) for the ith model target state of the ith class at the moment of k-1Performing one-step transfer to obtain the ith type jth model target prediction state
Wherein,andrespectively obtaining an ith model target motion prediction state and a target expansion prediction state;is in the ith class of jth modelAnda one-to-one mapping function ofji(. is) in the ith class of jth modelsAnda one-to-one mapping function of;
(4b) predicting the state according to the ith model targetAnd measuring the target collected at the moment k to obtain the target updating state under the correlated event l at the moment kAnd corresponding model likelihood
Wherein,representing n in time k associated events lkMeasured dispensing result,. phi(ji)l(. v) is associated with event l in the jth model of class iAnda one-to-one mapping function of phi(ji)l(. v) class i model under a correlation event /)Andis a one-to-one mapping function of f (·,) isAnda one-to-one mapping function of;andrespectively updating a motion state and an extended state of a target under the ith type jth model association event l,model likelihood under the correlation event l in the ith model and the jth model;
(4c) updating the state of the target obtained under all the associated eventsAnd corresponding model likelihoodCalculating to obtain the target update state of the ith class jth model at the moment k
Wherein,the motion state is updated for the target of the ith class jth model,updating an extended state for the target of the ith type jth model;
(4d) according to model likelihood of ith class jth model under correlation event lWith associated event numberObtaining model likelihood of ith class jth model
3. The method according to claim 1, wherein the model probability of the jth model of class i at time k is obtained in step (5)Determined by the following formula:
wherein, ckIs a normalization constant.
4. The method of claim 1, wherein the similarity of class i at time k is calculated in step (7)Determined by the following formula:
wherein,is the model likelihood of the ith class jth model,is the model probability of the ith class jth model.
5. The method of claim 1, wherein the target state of the ith class is calculated in step (6)Determined by the following formula:
wherein,is the model probability of the ith class jth model at time k,the motion state of the target after state updating in the ith model and the jth model,the updated target extension state in the ith model,is the moving state of the object in the ith class,target extension state in class i.
6. The method of claim 1, wherein the class i probability at time k is calculated in step (7)Determined by the following formula:
wherein,ncis a number of the classes,is the probability of class i at time k-1,is similar to the ith class at time k.
7. The method according to claim 1, wherein the motion states of all sub-ellipses in the i-th class are obtained in the step (8)And extended stateAnd calculating a motion state estimate of the sub-ellipse sAnd extended state estimationIs determined by the following steps;
(7a) according to motion state in class iAnd extended stateCalculating the motion state of the main ellipse and the sub ellipse in the ith classAnd extended state
Wherein λ is a constant;
(7b) according to the motion state of main and sub-ellipses in the ith classAnd extended stateCalculating the motion state of all the sub-ellipses in the ith class by using the non-ellipse structure information contained in the ith classAnd extended state
Wherein d iss,iThe coordinates of the sub-ellipse s representing the ith class with the main sub-ellipse as a reference point,andrespectively, major and minor ellipse expansion statesA rotation matrix and a diagonal matrix obtained by singular value decomposition, the diagonal matrixThe middle diagonal elements are arranged in descending order of the square of the semi-axis of the main ellipse; diagonal matrixThe middle diagonal element is the ratio of the sub-ellipse s of the ith class to the semi-axis of the main sub-ellipse, the rotation matrixIs the direction of the sub-ellipse s of the non-elliptical structure of the i-th class relative to the main sub-ellipse (·)TRepresenting a transpose operation on a matrix;
(7c) according to the motion state of all sub-ellipses in the ith classExtended stateAnd probability of class iCalculating to obtain motion state estimation of sub-ellipse sAnd extended state estimation:
where E [. cndot. ] is the averaging operation.
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 | |
| CN103729859B (en) | A kind of probability nearest neighbor domain multi-object tracking method based on fuzzy clustering | |
| CN103235933B (en) | A kind of vehicle abnormality behavioral value method based on HMM | |
| WO2017124299A1 (en) | Multi-target tracking method and tracking system based on sequential bayesian filtering | |
| CN107462882B (en) | Multi-maneuvering-target tracking method and system suitable for flicker noise | |
| CN107066806B (en) | Data Association and device | |
| CN110187337B (en) | LS and NEU-ECEF space-time registration-based high maneuvering target tracking method and system | |
| CN115731268A (en) | Unmanned aerial vehicle multi-target tracking method based on visual/millimeter wave radar information fusion | |
| WO2018227491A1 (en) | Method and device for association of fuzzy data of multiple targets in video | |
| WO2022021661A1 (en) | Gaussian process-based visual positioning method, system, and storage medium | |
| CN104237862B (en) | Probability hypothesis density filter radar system error fusion estimation method based on ADS-B | |
| CN114170274A (en) | Target tracking method and device, electronic equipment and storage medium | |
| CN112325885A (en) | Factor graph co-location algorithm based on mathematical statistical characteristics | |
| CN104777465B (en) | Random extended object shape and state estimation method based on B spline function | |
| CN113822335A (en) | GPB 1-GM-PHD-based sequential fusion target tracking method | |
| CN109214432A (en) | A kind of multiple-sensor and multiple-object joint-detection, tracking and classification method | |
| Xu et al. | Dynamic vehicle pose estimation and tracking based on motion feedback for LiDARs | |
| CN114049767B (en) | Edge computing method and device and readable storage medium | |
| CN108320302B (en) | CBMeMBer multi-object tracking method based on random hypersurface | |
| CN107391446A (en) | Irregular shape based on random matrix extends target shape and method for estimating state more | |
| CN113511194A (en) | Longitudinal collision avoidance early warning method and related device | |
| Chen et al. | Aerial robots on the way to underground: An experimental evaluation of VINS-mono on visual-inertial odometry camera | |
| CN116047495B (en) | State transformation fusion filtering tracking method for three-coordinate radar | |
| CN114415157B (en) | Underwater target multi-model tracking method based on underwater acoustic sensor network | |
| CN115471526A (en) | Automatic driving target detection and tracking method based on multi-source heterogeneous information fusion |
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 |