CN110516193A - Maneuvering target tracking method based on conversion Rayleigh filter under cartesian coordinate system - Google Patents
Maneuvering target tracking method based on conversion Rayleigh filter under cartesian coordinate system Download PDFInfo
- Publication number
- CN110516193A CN110516193A CN201910610729.3A CN201910610729A CN110516193A CN 110516193 A CN110516193 A CN 110516193A CN 201910610729 A CN201910610729 A CN 201910610729A CN 110516193 A CN110516193 A CN 110516193A
- Authority
- CN
- China
- Prior art keywords
- model
- acceleration
- filtering
- formula
- rayleigh
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Abstract
The present invention relates to the maneuvering target tracking methods based on conversion Rayleigh filter under a kind of cartesian coordinate system.Present invention relates particularly to the maneuvering target track questions of only angle measurement information.The present invention includes two parts, is to use the interactive multi-model centralization conversion Rayleigh proposed by the present invention with amendment acceleration to filter tracking of (MAIMMCSRF) the algorithm realization to maneuvering target based on the determination of measurement equation under the cartesian coordinate for converting Rayleigh filter thought and according to the measurement equation of cartesian coordinate system respectively.Core of the invention is the tracking based on the MAIMMCSRF algorithm proposed under the generation and cartesian coordinate system of measurement equation under the cartesian coordinate for converting Rayleigh filter thought to maneuvering target.The present invention to highly maneuvering target and can realize more accurate tracking and positioning, reduce acceleration error in interacting multiple model algorithm using median filtering in maneuvering Target Tracking Algorithm, target following error of the present invention is smaller, and tracking and positioning is more acurrate.
Description
Technical field
The invention belongs to sensor technical fields, and in particular to the only Target Tracking Problem of angle measurement.
Background technique
Radar is current most effective electronic remote detecting devices, it is irradiated and connects to target by emitting electromagnetic wave
Withdraw wave, thus to obtain target position and attribute information to achieve the purpose that positioning, tracking and the identification to target.It is collecting
When target echo carries out detection target and positions to it, radar needs to lead to its transmitting to powerful electromagnetic wave is radiated in the air
Signal is easy to be found and intercepted and captured by enemy, thus is subject to the attack of electronic interferences and antiradiation missile.Increasingly emphasizing
Concealed attack and the today killed firmly, passive location technology have become the mainstream of localization method development.Not with active location technology
Together, passive location technology is due to not active transmission of electromagnetic signals, but the transmitting signal by receiving radiation determines the spoke
Penetrate the position in source.Have the advantages that operating distance is remote, good concealment, strong antijamming capability, beats for hidden detection and accurately
It hits and provides important means.In modern practical battlefield surroundings, obtainable very limited about clarification of objective data, mesh
Target azimuth almost becomes unique reliable metrical information.
Usual passive tracking system can only obtain azimuth/pitch angle information of target, belong to incomplete observation.Only angle
Single station ESM sensor of degree information is not easily accomplished target following, there are problems that weak observability.Currently, being sensed for ESM
The research of device Target Tracking Problem is concentrated mainly on two aspects of selection of filtering algorithm and coordinate system.Due to only having angle measurement
One nonlinear estimation problem of Target Tracking Problem essence above formula, thus can only using nonlinear filtering algorithm carry out recursion with
Track.Traditional non-linear filtering method is Extended Kalman filter (EKF), it by the Taylor expansion of nonlinear function into
The truncation of row first-order linearization, to convert nonlinear problem to linearly, locating and tracking effect is estimated dependent on original state
Meter, easily there is morbid state in the corresponding covariance that filters, so as to cause the Divergent Phenomenon of filter.Although in order to make up this not
Foot, there is numerous improved methods to EKF, as EKF, iteration EKF etc., but the still difficult thoroughly solution of the above problem is truncated in high-order.Due to
The probability density distribution of nonlinear function is easier than approximate non-linear function, and people consider to use method of sampling approximate non-linear
Distribution is to solve nonlinear problem, the nonlinear filtering algorithm based on sampling, wherein that representative is particle filter (PF)
Algorithm and Unscented kalman filtering (UKF) algorithm, although both algorithms yield good result, calculation amount is larger.
2005, Clark et al. proposed conversion Rayleigh filtering (SRF), which is utilized using the nonlinear organization of measurement equation
Projection of the n-dimensional vector on unit circle or unit sphere, under conditions of previous moment state Gaussian distributed, using current
The conditional probability density of target bearing is precisely calculated in the observation information at moment, and approximate so as to avoid other filtering is missed
Difference is a kind of particularly suitable for only having the new algorithm of the targets passive tracking in the case of angle measurement, and theoretical and emulation shows
Only in the case where angle measurement, positioning accuracy approaches PF, and calculation amount is much smaller than the latter.
Median filtering is a kind of special nonlinear filtering algorithm, while having the characteristic for inhibiting noise and Protect edge information,
In signal field by very big attention.It mainly has the advantages that two is big: (1) can filtering completely spike, (pulse length is small
In filter width), be conducive to eliminate the high-frequency noise in acceleration signal.It (2) can be by step and slope, for symmetrical
Wavelet, median filtering will not cause phase distortion.There are also deficiencies for it, are mainly reflected in signal distortion and (filter longer, signal
Distort more serious), it can introduce false high frequency.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides under a kind of cartesian coordinate system based on conversion Rayleigh filter
Maneuvering target tracking method.
The conversion Rayleigh filtering (SRF) that J.M.C.Clark is proposed is the tracking problem for the only angle measurement of 2/3 dimension
It is a kind of new at the time of matched filtering algorithm.
If the state equation and measurement equation of system are respectively as follows:
X (k)=FX (k-1)+Us(k-1)+W(k-1) (1)
B (k)=Π [HX (k)+Um(k)+V(k)] (2)
In formula: F is system transfer matrix；UsIt (k-1) is input signal known to system；H is to be augmented measurement matrix；Um(k) it is
Measure input signal；W (k-1), V (k) are independent zero intermediate value, covariance is respectively Qs(k-1)、Qm(k) white Gaussian noise；
Π indicates projection of the n-dimensional vector on unit circle (n=2) or unit sphere (n=3).As n=2, the measurement of acquisition only has side
Parallactic angle θ (k), then B (k)=[sin (θ (k)), cos (θ (k))]T；As n=3, the measurement of acquisition has azimuth angle theta (k) and pitching
Angle η (k), then
B (k)=[cos (θ (k)) cos (η (k)), sin (θ (k)) cos (η (k)), sin (η (k))]T。
In the present invention, the state equation and measurement equation of system be still be (1), (2) formula, but Π is indicated in formula (2)
To be n-dimensional vector project in two dimension or three Cartesian coordinates.As n=2, the measurement of acquisition only has azimuth angle theta (k),
Then B (k)=[sin (θ (k)), cos (θ (k))]T；As n=3, the measurement of acquisition has azimuth angle theta (k) and pitch angle η (k), then
B (k)=[sin (θ (k)) cos (η (k)), sin (θ (k)) cos (η (k)), sin (η (k))]T。
In the tracking of maneuvering target, moving not enough, only in different fortune for target is only described by a model
The movement of target could be described preferably under dynamic state using different motion models.Interactive multi-model is mentioned based on the thought
Out, this method realizes the tracking to target using parallel multiple model filters, and when motor driven occurs for target, energy
Enough that adaptive " soft handover " is realized between a model, the final state estimation of target is that its each moment difference model must be estimated
Count the weighted blend of result.Interacting multiple algorithm is the target tracking algorism under a kind of motor-driven situation suitable for object height,
It is said from global angle, its tracking effect is best.But even with interacting multiple model filters, acceleration when target maneuver
There are still large errors.The present invention proposes that interactive multi-model, centralized Rayleigh filtering are formed in conjunction with median filtering and had
Correct the interactive multi-model centralization conversion Rayleigh filtering (MAIMMCSRF) of acceleration.This algorithm reduces interactive multimode
The error of acceleration estimation, improves tracking accuracy in type algorithm.The present invention proposes MAIMMCSRF target tracking algorism, shifts onto
The expression formula and median filtering acceleration estimation process of its likelihood function, and then provide complete cyclic process.
The present invention is described from the cyclic process of interactive multi-model.Interactive multi-model recursion circulation is handed over by input
Mutually, model filtering, model probability update, 4 steps of interaction output and acceleration estimation form.It carries out needing to count when model filtering
It calculates the likelihood function of each model, that is, utilizes the previous moment Target state estimator of each mode input, the new sight of estimation current time
Measure btLikelihood function, with model(at the time of t indicates state estimation, j indicates the corresponding model of t moment) matches seemingly
Right function may be expressed as:
Model filtering is using centralization conversion Rayleigh filtering, it is assumed that carries out fusion tracking to target for M station by quantity, often
The observed quantity that a moment obtains is bi,t(i indicates observation station or observation platform, i=1,2,3 ..., M), then about bi,tLikelihood
Function are as follows:
Likelihood function estimation is carried out simultaneously to the observed quantity of all observation stations (M), is obtainedThe matched likelihood function of institute
Are as follows:
Wherein,Respectively in jth model filter observation station i it is corresponding it is new breath size and filtering residuals it is equal
Variance.Formula (5) is likelihood function calculation method.By interactive multi-model, centralized Rayleigh filtering shape in conjunction with acceleration estimation
At the interactive multi-model centralization conversion Rayleigh filtering (MAIMMCSRF) with amendment acceleration, specific circulation step is as follows,
Wherein observation station number is M, Number of Models N, and the measurement noise of observation station i is in jth model filterTarget movement
Mode input beIn formulaWithRespectively indicate x, y in model j cartesian coordinate system and
The acceleration in z-axis direction.It is P that model, which jumps probability,j={ pij}N×N, (j=1,2 ..., N；I=1,2 ..., M).
Step 1, input interaction
By previous moment, that is, t-1 moment each model filter Target state estimator(j indicate j-th of model) with
The model probability of each filterObtain hybrid estimationAnd covarianceAnd it is followed hybrid estimation as t moment
The original state of ring:
Step 2, model filtering
Based on the dbjective state initialized againAnd covarianceObtaining new measurement btLater, with concentration
Formula converts Rayleigh filter and carries out state-updating, and according to formula (21) computation model likelihood function:
In its formula (18)Be expressed as follows:
In formula (19), r indicates the dimension of scene, in formulaFor convert Rayleigh filtering transformation mean value:
WhereinWithReduced form it is as follows:
Wherein,For the aggregation function of (0,1) standardized normal distribution N, may be expressed as:
Wherein, mode input matrix G and process noise matrix QSNot are as follows:
Calculated using residual sum covariance withThe likelihood function to match.
Step 3, model probability update
The calculating of model likelihood function is carried out by formula (21), then t moment model probability updates are as follows:
Step 4, interaction output and acceleration correction
(a) interaction output
Based on t moment model probabilityMerging is weighted to the estimated result of each filter, after being merged
EstimationWith covariance Pt|t。
(b) acceleration correction
The amendment of acceleration uses median filtering in the present invention.By by X, Y and Z obtained in interacting multiple model algorithm
The acceleration information in direction, is denoted as respectivelyWithAnd use ax(t|t)、ay(t | t) and az(t|
T) acceleration after respectively indicating median filtering.Median filtering is carried out by taking the acceleration of modified chi axis direction as an example.Assuming that X-axis adds
Speed data is stated by following formula (28):
{ai(i=1, K, Nx) (28)
I indicates to obtain No. i-th acceleration information, N in formulaxIndicate the number of X-axis acceleration information.Assuming that given intermediate value
The length of filtering is N (sliding window length, generally odd number), then to the median filtering of the K point of the i-th number are as follows:
(1) N number of sampled point is taken, centered on k-th point；
(2) to this N number of sampled point to sort from small to large；
(3) output valve of the center point value of N number of sampled point after taking sequence as K.
Special circumstances:
{aiIn acceleration information number be less than number of sampling points N when, it is assumed that N=5 then has
Work as NxWhen=1, then there is ax(t | t)=a1, wherein
Work as NxWhen=2, then have
Work as NxWhen=3, then have a1、a2WithIt arranges from small to large ord, takes median；
Work as NxWhen=4, first by { aiWithBy arranging from small to large, two values in centre are averaged.
Work as NxWhen >=5, a is acquired according to the method for above-mentioned median filteringxThe value of (t | t).
It repeats the above process, a can be obtainedx(t|t)、ay(t | t) and az(t | t), to modify and update model parameter.
The invention has the advantages that:
The tracking of maneuvering target uses interactive multi-model process in the present invention, and is calculated using the filtering of newest conversion Rayleigh
Method carries out optimal estimation to the location information of maneuvering target, and estimates to make up interactive multi-model to maneuvering target acceleration
The problem of Accuracy extimate deficiency is counted, the present invention is modified estimation to acceleration using the method for median filtering.
Specific embodiment
Specific implementation of the invention mainly includes three parts, respectively the acquisition of the observation information of observation station, non-maneuver mesh
Target tracking, the tracking of maneuvering target.
1. the acquisition of observation information
Assuming that observation station number is M, measurement dimension is n.As n=2, then the M measurement information that can be measured from observation station is i.e.
Azimuth angle theta (k), corresponding measurement equation are B (k)=[sin (θ (k)), cos (θ (k))]T；It, then can be from observation as n=3
Station measures M to measurement information i.e. azimuth angle theta (k) and pitch angle η (k), and corresponding measurement equation is B (k)=[sin (θ (k))
cos(η(k)),sin(θ(k))cos(η(k)),sin(η(k))]T.So far observation station can be obtained and target is formed by vector
Projection in cartesian coordinate system.In addition, there is angle error σ and system noise covariance Q in each observation stations。
2. maneuvering target tracking
In the tracking of maneuvering target, moving not enough, only in different fortune for target is only described by a model
The movement of target could be described preferably under dynamic state using different motion models.This method is using parallel multiple models filter
Wave device realizes the tracking to target, and when motor driven occurs for target, adaptively " soft to cut can be realized between a model
Change ", the final state estimation of target is the weighted blend that its each moment difference model obtains estimated result.Interactive Multiple-Model is calculated
Method is the target tracking algorism under a kind of motor-driven situation suitable for object height, is said from global angle, its tracking effect is best.
But filtered when target maneuver even with Interactive Multiple-Model, there are still large errors for acceleration.The present invention is more by interactive mode
Model, centralized Rayleigh filtering form auspicious with the interactive multi-model centralization conversion for correcting acceleration in conjunction with median filtering
Benefit filtering (MAIMMCSRF).This algorithm reduces the error of acceleration estimation in interacting multiple model algorithm, improves tracking essence
Degree.The algorithm the specific implementation process is as follows.
Step 1: input interaction.
By previous moment, that is, t-1 moment each model filter Target state estimator(j indicate j-th of model) with
The model probability of each filterObtain hybrid estimationAnd covarianceAnd it is followed hybrid estimation as t moment
The original state of ring:
When initial time, the Target state estimator of initial each model given firstAnd corresponding model probabilityAssuming that observation station number be M, Number of Models N, target movement mode input beIn formulaWithRespectively indicate the acceleration in the x, y and z axes direction in model j cartesian coordinate system.It is P that model, which jumps probability,j
={ pij}N×N, (j=1,2 ..., N；I=1,2 ..., M).
Step 2: model filtering.
Measurement noise by observation station i in 1 available jth model filter isSystem noise covariance QsAnd
It obtains measuring vector from measurement equation B (k).Based on the dbjective state initialized againCovarianceIt is shifted with system
Matrix F carries out state-updating with centralization conversion Rayleigh filter, and according to formula (21) computation model likelihood function:
Wherein, mode input matrix are as follows:Process noise matrix are as follows:
Calculated using residual sum covariance withThe likelihood function to match.It is implemented as follows:
According to formula (5) can be calculated withThe likelihood function to match.
Step 3: model probability updates.
The calculating of model likelihood function is carried out by formula (21), then t moment model probability updates are as follows:
Step 4: interaction output and acceleration correction.
(a) interaction output
Based on t moment model probabilityMerging is weighted to the estimated result of each filter, after being merged
EstimationWith covariance Pt|t。
(b) acceleration correction
The amendment of acceleration uses median filtering in the present invention.By by X, Y and Z obtained in interacting multiple model algorithm
The acceleration information in direction, is denoted as respectivelyWithAnd use ax(t|t)、ay(t | t) and az(t|
T) acceleration after respectively indicating median filtering.Median filtering is carried out by taking the acceleration of modified chi axis direction as an example.Assuming that X-axis adds
Speed data is stated by following formula (28):
{ai(i=1, K, Nx) (28)
I indicates to obtain No. i-th acceleration information, N in formulaxIndicate the number of X-axis acceleration information.Assuming that given intermediate value
The length of filtering is N (sliding window length, generally odd number), then to the median filtering of the K point of the i-th number are as follows:
(1) N number of sampled point is taken, centered on k-th point；
(2) to this N number of sampled point to sort from small to large；
(3) output valve of the center point value of N number of sampled point after taking sequence as K.
Special circumstances:
{aiIn acceleration information number be less than number of sampling points N when, it is assumed that N=5 then has
Work as NxWhen=1, then there is ax(t | t)=a1, wherein
Work as NxWhen=2, then have
Work as NxWhen=3, then have a1、a2WithIt arranges from small to large ord, takes median；
Work as NxWhen=4, first by { aiWithBy arranging from small to large, two values in centre are averaged.
Work as NxWhen >=5, a is acquired according to the method for above-mentioned median filteringxThe value of (t | t).
It repeats the above process, a can be obtainedx(t|t)、ay(t | t) and az(t | t), to modify and update model parameter.
Based on above four steps, tracking of the method proposed in the present invention to maneuvering target, tracking essence may be implemented
Degree is higher than traditional interactive multi-model.
Claims (1)
1. the maneuvering target tracking method based on conversion Rayleigh filter under cartesian coordinate system, it is characterised in that:
The state equation and measurement equation of system are respectively as follows:
X (k)=FX (k-1)+Us(k-1)+W(k-1) (1)
B (k)=Π [HX (k)+Um(k)+V(k)] (2)
In formula: F is system transfer matrix；UsIt (k-1) is input signal known to system；H is to be augmented measurement matrix；UmIt (k) is measurement
Input signal；W (k-1), V (k) are independent zero intermediate value, covariance is respectively Qs(k-1)、Qm(k) white Gaussian noise, Π n
Dimensional vector projects in two dimension or three Cartesian coordinates；As n=2, the measurement of acquisition only has azimuth angle theta (k), then B (k)
=[sin (θ (k)), cos (θ (k))]T；As n=3, the measurement of acquisition has azimuth angle theta (k) and pitch angle η (k), then B (k)=
[sin(θ(k))cos(η(k)),sin(θ(k))cos(η(k)),sin(η(k))]T
This method is described from the cyclic process of interactive multi-model；Interactive multi-model recursion circulation is by input interaction, mould
Type filtering, model probability update, four steps of interaction output and acceleration estimation form；It carries out needing to calculate when model filtering each
The likelihood function of model utilizes the previous moment Target state estimator of each mode input, the new observed quantity of estimation current time
btLikelihood function, with modelThe likelihood function to match may be expressed as:
Model filtering is using centralization conversion Rayleigh filtering, it is assumed that carries out fusion tracking, Mei Geshi to target for M station by quantity
Carving obtained observed quantity is bi,t, then about bi,tLikelihood function are as follows:
Likelihood function estimation is carried out simultaneously to the observed quantity of all observation stations, is obtainedThe matched likelihood function of institute are as follows:
Wherein,The mean square deviation of observation station i corresponding new breath size and filtering residuals respectively in jth model filter；
Formula (5) is likelihood function calculation method；Interactive multi-model, centralized Rayleigh filtering are formed in conjunction with acceleration estimation and had
The interactive multi-model centralization conversion Rayleigh filtering (MAIMMCSRF) of acceleration is corrected, specific circulation step is as follows: wherein seeing
Survey station number is M, Number of Models N, and the measurement noise of observation station i is in jth model filterThe model of target movement is defeated
Enter forIn formulaWithRespectively indicate the x, y and z axes direction in model j cartesian coordinate system
Acceleration；It is P that model, which jumps probability,j={ pij}N×N:
Step 1, input interaction
By previous moment, that is, t-1 moment each model filter Target state estimatorWith the model probability of each filterObtain hybrid estimationAnd covarianceAnd the original state for recycling hybrid estimation as t moment:
Step 2, model filtering
Based on the dbjective state initialized againAnd covarianceObtaining new measurement btLater, it is converted with centralization
Rayleigh filter carries out state-updating, and according to formula (21) computation model likelihood function:
In its formula (18)Be expressed as follows:
In formula (19), r indicates the dimension of scene, in formulaFor convert Rayleigh filtering transformation mean value:
WhereinWithReduced form it is as follows:
Wherein,For the aggregation function of (0,1) standardized normal distribution N, may be expressed as:
Wherein, mode input matrix are as follows:Process noise matrix are as follows:
Calculated using residual sum covariance withThe likelihood letter to match
Number；
Step 3, model probability update
The calculating of model likelihood function is carried out by formula (21), then t moment model probability updates are as follows:
Step 4, interaction output and acceleration correction
4-1, interaction output
Based on t moment model probabilityMerging is weighted to the estimated result of each filter, the estimation after being mergedWith covariance Pt|t；
4-2, acceleration correction
The amendment of acceleration uses median filtering；By by the acceleration of X, Y obtained in interacting multiple model algorithm and Z-direction
Information is denoted as respectivelyWithAnd use ax(t|t)、ay(t | t) and az(t | t) respectively indicate intermediate value
Filtered acceleration；Median filtering is carried out by taking the acceleration of modified chi axis direction as an example；Assuming that X-axis acceleration information is by following formula
(28) it states:
{ai(i=1, K, Nx) (28)
I indicates to obtain No. i-th acceleration information, N in formulaxIndicate the number of X-axis acceleration information；Assuming that given median filtering
Length be N, then to the median filtering of the K point of the i-th number are as follows:
(1) N number of sampled point is taken, centered on k-th point；
(2) to this N number of sampled point to sort from small to large；
(3) output valve of the center point value of N number of sampled point after taking sequence as K；
{aiIn acceleration information number be less than number of sampling points N when, it is assumed that N=5 then has
Work as NxWhen=1, then there is ax(t | t)=a1, wherein
Work as NxWhen=2, then have
Work as NxWhen=3, then have a1、a2WithIt arranges from small to large ord, takes median；
Work as NxWhen=4, first by { aiWithBy arranging from small to large, two values in centre are averaged；
Work as NxWhen >=5, a is acquired according to the method for above-mentioned median filteringxThe value of (t | t)；
It repeats the above process, a can be obtainedx(t|t)、ay(t | t) and az(t | t), to modify and update model parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910610729.3A CN110516193B (en) | 2019-07-08 | 2019-07-08 | Maneuvering target tracking method based on transformation Rayleigh filter under Cartesian coordinate system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910610729.3A CN110516193B (en) | 2019-07-08 | 2019-07-08 | Maneuvering target tracking method based on transformation Rayleigh filter under Cartesian coordinate system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110516193A true CN110516193A (en) | 2019-11-29 |
| CN110516193B CN110516193B (en) | 2023-03-21 |
Family
ID=68623808
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910610729.3A Active CN110516193B (en) | 2019-07-08 | 2019-07-08 | Maneuvering target tracking method based on transformation Rayleigh filter under Cartesian coordinate system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110516193B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111416595A (en) * | 2020-04-08 | 2020-07-14 | 北京航空航天大学 | Big data filtering method based on multi-core fusion |
| CN111736146A (en) * | 2020-07-03 | 2020-10-02 | 哈尔滨工业大学 | Bistatic pre-detection tracking method and device based on speed filtering |
| CN113447919A (en) * | 2021-06-29 | 2021-09-28 | 重庆大学 | Extended Kalman prediction angle tracking method |
| CN113963025A (en) * | 2021-10-22 | 2022-01-21 | 西北工业大学深圳研究院 | Underwater self-adaptive maneuvering target rapid tracking and tracing method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107045125A (en) * | 2017-03-17 | 2017-08-15 | 电子科技大学 | A kind of Interactive Multiple-Model radar target tracking method based on predicted value measurement conversion |
| CN107193009A (en) * | 2017-05-23 | 2017-09-22 | 西北工业大学 | A kind of many UUV cooperative systems underwater target tracking algorithms of many interaction models of fuzzy self-adaption |
| CN108896986A (en) * | 2018-04-23 | 2018-11-27 | 电子科技大学 | A kind of measurement conversion Sequential filter maneuvering target tracking method based on predicted value |
| CN109001699A (en) * | 2018-01-30 | 2018-12-14 | 哈尔滨工业大学 | Based on the tracking constrained with noise destination information |
-
2019
- 2019-07-08 CN CN201910610729.3A patent/CN110516193B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107045125A (en) * | 2017-03-17 | 2017-08-15 | 电子科技大学 | A kind of Interactive Multiple-Model radar target tracking method based on predicted value measurement conversion |
| CN107193009A (en) * | 2017-05-23 | 2017-09-22 | 西北工业大学 | A kind of many UUV cooperative systems underwater target tracking algorithms of many interaction models of fuzzy self-adaption |
| CN109001699A (en) * | 2018-01-30 | 2018-12-14 | 哈尔滨工业大学 | Based on the tracking constrained with noise destination information |
| CN108896986A (en) * | 2018-04-23 | 2018-11-27 | 电子科技大学 | A kind of measurement conversion Sequential filter maneuvering target tracking method based on predicted value |
Non-Patent Citations (3)
| Title |
|---|
| 刘梅;陈锦海;高扬;王骏;林超;张声杰;: "基于IMM-CSRF的多平台机动目标被动跟踪方法" * |
| 张恒;高敏;徐超;: "改进的强跟踪容积卡尔曼滤波的机动目标跟踪" * |
| 潘平俊;冯新喜;李锋;石磊;李鸿艳;: "基于转换瑞利滤波器的红外目标跟踪算法" * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111416595A (en) * | 2020-04-08 | 2020-07-14 | 北京航空航天大学 | Big data filtering method based on multi-core fusion |
| CN111416595B (en) * | 2020-04-08 | 2022-04-08 | 北京航空航天大学 | Big data filtering method based on multi-core fusion |
| CN111736146A (en) * | 2020-07-03 | 2020-10-02 | 哈尔滨工业大学 | Bistatic pre-detection tracking method and device based on speed filtering |
| CN111736146B (en) * | 2020-07-03 | 2022-06-21 | 哈尔滨工业大学 | Bistatic pre-detection tracking method and device based on speed filtering |
| CN113447919A (en) * | 2021-06-29 | 2021-09-28 | 重庆大学 | Extended Kalman prediction angle tracking method |
| CN113963025A (en) * | 2021-10-22 | 2022-01-21 | 西北工业大学深圳研究院 | Underwater self-adaptive maneuvering target rapid tracking and tracing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110516193B (en) | 2023-03-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110516193A (en) | Maneuvering target tracking method based on conversion Rayleigh filter under cartesian coordinate system | |
| CN105929378B (en) | Combine the direct tracking of time delay and Doppler frequency based on external sort algorithm | |
| CN107193009A (en) | A kind of many UUV cooperative systems underwater target tracking algorithms of many interaction models of fuzzy self-adaption | |
| CN106443661A (en) | Maneuvering extended target tracking method based on unscented Kalman filter | |
| CN108761387B (en) | Double-station time difference and frequency difference combined positioning method for fixed radiation source | |
| CN104777478A (en) | Target searching and acquisition method of phased array radar | |
| CN104182609B (en) | The three-dimensional target tracking method that unbiased transformation based on decorrelation is measured | |
| CN109298725A (en) | A kind of Group Robots distributed multiple target tracking method based on PHD filtering | |
| CN115731268A (en) | Unmanned aerial vehicle multi-target tracking method based on visual/millimeter wave radar information fusion | |
| CN107436427A (en) | Space Target Motion Trajectory and radiation signal correlating method | |
| CN113342059B (en) | Multi-unmanned aerial vehicle tracking mobile radiation source method based on position and speed errors | |
| CN112146648A (en) | Multi-target tracking method based on multi-sensor data fusion | |
| CN103064084A (en) | Ambiguity solving method based on distance frequency domain | |
| CN108763158A (en) | Frequency difference combined calculation method and system when a kind of | |
| CN103499809B (en) | A kind of Pure orientation double computer cooperation target following location path planing method | |
| CN116794649A (en) | Clutter maneuvering target tracking method based on waveform selection | |
| CN206292389U (en) | A kind of alignment system of mobile target | |
| CN113985376B (en) | Radar comprehensive display and control excitation system | |
| CN113933876B (en) | Multi-star communication time difference positioning data fusion processing method | |
| CN115079159A (en) | Airborne radar motion error compensation and accumulation detection method based on IGRFT | |
| CN107329117A (en) | It is a kind of that compensation method is composed based on the bistatic airborne radar self-adapting clutter for improving OMP | |
| CN112666516A (en) | Passive tracking method based on track information field | |
| CN106066471A (en) | The localization method of a kind of mobile target and system | |
| CN109934840A (en) | Circumference SAR motion target tracking method based on GMPHD filter | |
| Zhang et al. | The non-linear tracking of IMM-PHD filter for rader-infrared sensor data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |