CN109901153A - Targetpath optimization method based on information entropy weight and nearest-neighbor data correlation - Google Patents
Targetpath optimization method based on information entropy weight and nearest-neighbor data correlation Download PDFInfo
- Publication number
- CN109901153A CN109901153A CN201910248309.5A CN201910248309A CN109901153A CN 109901153 A CN109901153 A CN 109901153A CN 201910248309 A CN201910248309 A CN 201910248309A CN 109901153 A CN109901153 A CN 109901153A
- Authority
- CN
- China
- Prior art keywords
- moment
- measuring point
- target
- measurement
- vector
- 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
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses the targetpath optimization methods based on information entropy weight and nearest-neighbor data correlation, comprising the following steps: the measurement vector for obtaining the first two moment measuring point after track initiation calculates the initial state vector of target；Kalman filter is initialized；State estimation vector, evaluated error covariance matrix, predicted vector, new breath covariance matrix and the kalman gain of k moment target are calculated using iterative method；The acquisition k moment falls into the candidate measuring point of center of tracking gate；When the number of the candidate measuring point at k moment is greater than 1, update is optimized using targetpath of the entropy assessment to the k moment.The present invention excavates known measurement information by depth, on the basis of not changing prior information, takes full advantage of target and measures entrained information, have lower improvement cost and better applicability；Meanwhile improving the accuracy of radar tracking track.
Description
Technical field
The invention belongs to Radar Technology field more particularly to a kind of mesh based on information entropy weight and nearest-neighbor data correlation
Mark route optimization method.
Background technique
Currently, continuing to bring out with various new system radars, the utilizable characteristic in multiple target tracking field is also more next
More diversification, and data association technique remains the key technology in multiple target tracking field, data correlation is in target and its amount
Between survey according to specific data association algorithm establish a kind of mapping relations, can be judged with data correlation target specifically with
Which, which is measured, is associated.Therefore, the accuracy of data correlation seriously affects the performance of target tracking of radar.
Typical data correlation target tracking algorism has nearest-neighbor algorithm (nearest neighborhood, NN), general
Rate data association algorithm (probabilitY data association, PDA), Joint Probabilistic Data Association algorithm (joint
Probability data association, JPDA) and multiple hypotheis tracking algorithm (multiple hYpothesis
Tracking, MHT).Wherein, the working principle of nearest-neighbor algorithm is that tracking gate is first arranged, obtained by tracking gate preliminary screening
The echo arrived becomes candidate echo, to limit the target numbers for participating in correlation discriminating.Since this method principle is concise, calculate complicated
Lower, the characteristics of being easy to Project Realization is spent, in engineer application, for monotrack or high s/n ratio under low signal-to-noise ratio environment
Multiple target tracking under environment, nearest-neighbor algorithm, which suffers from, to be widely applied.
But existing nearest-neighbor algorithm is there are still data correlation accuracy is high, and filter result is inaccurate and multiple target
The problem of erroneous association is also easy to produce when tracking；Simultaneously as all properties equably participate in calculating during data correlation,
The significance level of not prominent attribute, the as a result influence vulnerable to single attribute.
Existing various types of improved method regarding to the issue above such as uses estimated location and subsequent time amount
Location set between the actual range method that replaces statistical distance, by introducing the method for radial doppler velocity information and passing through
The method etc. of measured value and map feature association probability is introduced, the accuracy promotion of data correlation is limited, and above method is all
It is to improve nearest-neighbor algorithm by introducing various new characteristic informations to realize, and introduce new characteristic information and will increase work
Cheng Shixian difficulty and improvement cost limit the popularization scope of application of improved method.
Summary of the invention
To solve the above-mentioned problems, the purpose of the present invention is to propose to one kind to be based on information entropy weight and nearest-neighbor data correlation
Targetpath optimization method (nearest neighbor based on information entropY weight,
IEWNN), known measurement information is excavated by depth and target is more fully utilized on the basis of not changing prior information
Information entrained by measuring point has lower improvement cost and better applicability；Meanwhile utilizing the information of target measuring point
Entropy improves the accuracy of radar track tracking.
According to the basic principle of information theory, information is a measurement of system order degree, and entropy is the unordered degree of system
One measurement.The average degree of indeterminacy of information source, information source output is random quantity in information theory, thus its degree of indeterminacy can use probability point
Cloth is measured.If the comentropy of index is smaller, the information content which provides is bigger, acts on and getting over played in overall merit
Greatly, i.e., weight is higher.
If source symbol has n kind value: U1, U2..., Un, corresponding probability are as follows: P1, P2..., Pn, and the appearance of various symbols
Independently of one another.At this moment, the average uncertainty of information source should be single symbol uncertainty-logPiAssembly average (E),
It can be described as comentropy
The entropy assessment drawn by comentropy can be realized the multipair overall merit as, multi objective, and the present invention utilizes information
Entropy accurately carries out tax power to the attribute of data object, improves the accuracy of Radar Target Track tracking.
In order to achieve the above object, the present invention is resolved using following technical scheme.
Targetpath optimization method based on information entropy weight and nearest-neighbor data correlation, comprising the following steps:
Step 1, the state equation of radar tracking system is set as Xk+1=FXkMeasurement equation with radar tracking system is Zk
=HXk+Wk；
The measuring point that the first two moment radar tracking system after setting track initiation obtains is unique substantial amount measuring point；
The measurement vector Z of the measuring point at the first two moment after the track initiation obtained using radar tracking system0, Z1, calculate target
Initial state vector
Wherein, Xk+1For the state vector of k+1 moment radar tracking system, XkFor k moment radar tracking system state to
Amount, WkFor the measurement noise sequence at k moment；F is the state-transition matrix of radar tracking system, and H is the amount of radar tracking system
Survey matrix；x0For the position of target X-direction under initial time two-dimensional Cartesian coordinate system, y0For initial time two-dimensional Cartesian coordinate system
The position of lower target Y-direction,For the speed of target X-direction under initial time two-dimensional Cartesian coordinate system,For initial time two
Tie up the speed of target Y-direction under rectangular coordinate system.
Step 2, according to the measurement vector Z of the measuring point at the first two moment after the track initiation of radar tracking system acquisition0,
Z1, Kalman filter is initialized by two o'clock calculus of finite differences, calculate the Initial state estimation of Kalman filter to
AmountWith initial estimation error co-variance matrix P0；
Step 3, according to the Initial state estimation vector of Kalman filterWith initial estimation error co-variance matrix P0,
K (k=1,2...n, n ∈ N is calculated using iterative method+) moment target state estimation vectorThe estimation of k moment target
Error co-variance matrix Pk/k-1, k moment target predicted vectorThe new breath covariance matrix S of k moment targetkWith the k moment
The kalman gain K of targetk；
Step 4, the measurement point set W that the radar scanning of k moment obtains is obtainedk, by the predicted vector of the k moment targetPosition as tracking gate center, tracking gate threshold value is chosen, to the measurement point set W at k momentkTentatively sieved
Choosing, obtains the candidate measuring point at k moment.
Step 5, when the number of the candidate measuring point at k moment is 0, then the state estimation vector of the k moment targetWith the evaluated error covariance matrix P of the k moment targetk/k-1For course extrapolation.
When the number of the candidate measuring point at k moment is 1, then the measurement vector of the candidate measuring point at the k moment is used for
Track updates.
When the number of the candidate measuring point at k moment is greater than 1, optimized using targetpath of the entropy assessment to the k moment
It updates.
Described optimized using targetpath of the entropy assessment to the k moment is updated comprising following sub-step:
Sub-step 5.1, according to the l of a candidate measuring point of the jth (j=1,2 ..., m, m > 1) at k moment (l=1,
2 ..., n, n >=1) a measurement index ajlWith the predicted vector of k moment targetCalculate k moment j-th candidates measuring point
First of measurement index ajlWith first of measurement index a of the prediction measuring point i at tracking gate centerilBetween error amount
Δjl:
Δjl=| ajl-ail|。
Step 5.2, referred to according to the measurement corresponding with the prediction measuring point i at tracking gate center of k moment j-th candidates measuring point
Error amount Δ between markjlWith the quantity m of k moment candidate measuring point, the uncertainty S of first of k moment measurement index is obtainedl；
The uncertainty S of index is measured to first of the k momentlIt is normalized, obtains the weight system of first of k moment measurement index
Number αl。
Step 5.3, the weight coefficient α of index is measured according to first of the k momentl, when using statistical distance relation maxim to k
The statistical distance carved between prediction measuring point i and candidate measuring point carries out tax power processing, obtains the tax power statistical distance at k moment
IEWdIj, k；Statistical distance IEWd is weighed in the tax for choosing the k momentIj, kPreferred mesh of the corresponding candidate's measuring point as k moment when minimum
Scalar measuring point optimizes update to the targetpath at k moment.
Utilize the measurement vector z of k moment preferred target measuring pointk', the state of the k moment target after calculating optimization
Estimate vectorWith evaluated error covariance matrix Pk:
Pk=Pk/k-1-KkHPk/k-1；
According to the state estimation vector of the k moment target after optimizationWith the evaluated error covariance matrix P of targetk, to k
The state estimation vector of the target at+1 momentWith evaluated error covariance matrix Pk+1/kIt is predicted, obtains k+1 moment mesh
Target predicted vectorNew breath covariance matrix Sk+1With kalman gain Kk+1, step 3 is repeated to step 5, and iteration goes out the mesh
Target optimizes track.
Compared with prior art, the invention has the benefit that
(1) present invention excavates known measurement information by further depth, on the basis of not changing prior information, more
Information entrained by target measuring point adequately is utilized, there is lower improvement cost and better applicability, and be easier to use
In the transformation and upgrade of relevant device.
(2) present invention passes through the comentropy of measurement information known to analytical calculation, and respective attribute is determined using entropy assessment
Weight coefficient, tax power is carried out to statistical distance relation maxim, enables each index of target measuring point with self-contained letter
The size of breath amount carries out tax power, avoids the shortcomings that tracking result is vulnerable to single properties affect.
(3) present invention is not limited solely to improving and optimizating under two dimension measures, and is applied equally under multidimensional measurement information
Radar tracking track optimization, and as measurement information gets over diversification, the information excavated is also abundanter, improves Track In Track
Accuracy.
Detailed description of the invention
The present invention is described in further details in the following with reference to the drawings and specific embodiments.
Fig. 1 is that the process of the targetpath optimization method of the invention based on information entropy weight and nearest-neighbor data correlation is shown
It is intended to.
Fig. 2 is the operation stream of the targetpath optimization method of the invention based on information entropy weight and nearest-neighbor data correlation
Cheng Tu.
Fig. 3 is target plot obtained in the embodiment of the present invention, wherein icon ture indicates true track, NN table
Show that track obtained by conventional nearest-neighbor, IEWNN indicate track obtained by the method for the present invention.
Fig. 4 is in the embodiment of the present invention using conventional nearest-neighbor (NN) and of the invention based on information entropy weight arest neighbors
The target following comparison diagram in domain (IEWNN)；Wherein, Fig. 4 (a) is X-direction distance versus figure, and icon X-ture indicates true track
X-direction distance, NN indicates that based on the conventional resulting X-direction of nearest-neighbor, IEWNN indicates the method for the present invention apart from tracking result
Resulting X-direction is apart from tracking result；Fig. 4 (b) is X-direction velocity contrast figure, and icon X-ture indicates the X-direction of true track
Speed, NN are indicated based on the conventional resulting X-direction speed tracing of nearest-neighbor as a result, IEWNN indicates the resulting X of the method for the present invention
Speed is apart from tracking result；Fig. 4 (c) is Y-direction distance versus figure, and icon Y-ture indicates the Y-direction distance of true track, NN
Indicate based on the conventional resulting Y-direction of nearest-neighbor apart from tracking result, IEWNN indicate the resulting Y-direction of the method for the present invention away from
From tracking result；Fig. 4 (d) is Y-direction velocity contrast figure icon, and Y-ture indicates the Y-direction distance of true track, and NN indicates base
In the resulting Y-direction of conventional nearest-neighbor apart from tracking result, IEWNN indicates the resulting Y-direction distance tracking of the method for the present invention
As a result.
Fig. 5 is in the embodiment of the present invention using conventional nearest-neighbor (NN) and of the invention based on information entropy weight arest neighbors
The target following error comparison diagram in domain (IEWNN)；Fig. 5 (a) is X-direction range error comparison diagram；Fig. 5 (b) is X-direction speed mistake
Poor comparison diagram；Fig. 5 (c) is Y-direction range error comparison diagram；Fig. 5 (d) is Y-direction velocity error comparison diagram.
Fig. 6 is in the embodiment of the present invention using conventional nearest-neighbor (NN) and of the invention based on information entropy weight arest neighbors
The target following Monte Carlo Experiment root-mean-square error comparison diagram in domain (IEWNN)；Fig. 6 (a) is X-direction distance root mean square error pair
Than figure；Fig. 6 (b) is X-direction speed root-mean-square error comparison diagram；Fig. 6 (c) is Y-direction distance root mean square error comparison diagram；Fig. 6
It (d) is Y-direction speed root-mean-square error comparison diagram.
Specific embodiment
Embodiment of the present invention is described in detail below in conjunction with embodiment, but those skilled in the art will
It will be appreciated that following embodiment is merely to illustrate the present invention, and it is not construed as limiting the scope of the invention.
Referring to Figures 1 and 2, the present invention includes the following steps:
Step 1, the state equation of radar tracking system is set as Xk+1=FXkMeasurement equation with radar tracking system is Zk
=HXk+Wk。
The measuring point that the first two moment radar tracking system after setting track initiation obtains is unique substantial amount measuring point；
Using the measurement vector Z of the measuring point at the first two moment after radar tracking system acquisition track initiation0, Z1, calculate the first of target
Beginning state vector
Wherein, Xk+1For the state vector of k+1 moment radar tracking system, XkFor k moment radar tracking system state to
Amount, WkFor the measurement noise sequence at k moment；F is the state-transition matrix of radar tracking system, and H is the amount of radar tracking system
Survey matrix；x0For the position of target X-direction under initial time two-dimensional Cartesian coordinate system, y0For initial time two-dimensional Cartesian coordinate system
The position of lower target Y-direction,For the speed of target X-direction under initial time two-dimensional Cartesian coordinate system,For initial time
The speed of target Y-direction under two-dimensional Cartesian coordinate system.
Specifically, it in step 1, sets under two dimensional motion, state-transition matrix F, the thunder of the radar tracking system
Up to the measurement matrix H of tracking system, the measurement vector Z of the k moment targetkWith the amount at the first two moment after the track initiation
The measurement vector Z of measuring point0, Z1Expression formula be respectively as follows:
Zk=[ZX, kZY, k]T, Z0=[ZX, 0ZY, 0]T, Z1=[ZX, 1ZY, 1]T；
In above formula, []TFor the transposition of matrix, t is the sampling interval；ZX, kFor the k moment under two-dimensional Cartesian coordinate system target
The distance of X-direction；ZY, kFor the distance of k moment target Y-direction under two-dimensional Cartesian coordinate system；ZX, 1It is target the 1st moment
X-direction position measuring value, Z under the two-dimensional Cartesian coordinate system of acquisitionX, 0The two-dimentional rectangular co-ordinate obtained for target the 0th moment
It is lower X-direction position measuring value, ZY, 1For target, Y-direction position is measured under the two-dimensional Cartesian coordinate system that the 1st moment obtains
Value, ZY, 0For target under the two-dimensional Cartesian coordinate system that the 0th moment obtains Y-direction position measuring value.
Step 2, according to the measurement vector Z of the measuring point at the first two moment after the track initiation of radar tracking system acquisition0,
Z1, Kalman filter is initialized by two o'clock calculus of finite differences, calculate the Initial state estimation of Kalman filter to
AmountWith initial estimation error co-variance matrix P0。
The above-mentioned specific calculating such as following formula that Kalman filter is initialized by two o'clock calculus of finite differences:
Z0=[ZX, 0ZY, 0]T, Z1=[ZX, 1ZY, 1]T；
Wherein, r is to measure noise W0Variance,For Kalman filter original state under two-dimensional Cartesian coordinate system
When X-direction position estimation value,For X-direction velocity estimation when Kalman filter original state under two-dimensional Cartesian coordinate system
Value,For Y-direction position estimation value when Kalman filter original state under two-dimensional Cartesian coordinate system,It is two-dimentional straight
Y-direction velocity estimation value, Z when Kalman filter original state under angular coordinate systemX, 1For target the 1st moment obtain two
Tie up X-direction position measuring value under rectangular coordinate system, ZX, 0For target under the two-dimensional Cartesian coordinate system that the 0th moment obtains X-direction
Position measuring value, ZY, 1For target under the two-dimensional Cartesian coordinate system that the 1st moment obtains Y-direction position measuring value, ZY, 0For mesh
It is marked on Y-direction position measuring value under the two-dimensional Cartesian coordinate system of the 0th moment acquisition.
Step 3, according to the Initial state estimation vector of Kalman filterWith initial estimation error co-variance matrix P0,
Iterate to calculate out k (k=1,2 ... n, n ∈ N+) moment target state estimation vectorEvaluated error covariance matrix
Pk/k-1, predicted vectorNew breath covariance matrix SkWith kalman gain Kk:
Pk/k-1=FPk-1FT
Sk=HPk/k-1HT+Rk
Wherein,For the state estimation vector of k-1 moment target, Pk-1For the evaluated error covariance of k-1 moment target
Matrix, RkNoise covariance matrix is measured for the k moment,TFor the transposition of matrix,-1For inverse of a matrix.
As k=1,AsPk-1As P0, by above formula be iterated calculating can be obtained except initial time with
The state estimation vector of the target of outer any moment, evaluated error covariance matrix, predicted vector, new breath covariance matrix and
Kalman gain.
Step 4, the measurement point set W that the radar scanning of k moment obtains is obtainedk, by k moment target obtained in step 3
Predicted vectorTracking gate threshold value is chosen, to the measurement point set at k moment as tracking gate center in the position at place
WkPreliminary screening is carried out, the candidate measuring point at k moment is obtained.
Specifically, point set W is measuredkIn comprising substantial amount measuring point and the false measuring point that is randomly generated, will be in step 3
The predicted vector of the k moment target arrivedThe position at place sets the tracking of nearest-neighbor algorithm as tracking gate center
Wave door is chosen under two-dimensional Cartesian coordinate system according to elliptical wave door rule.To track the predicted position of target as center of tracking gate,
The design of wave door size, which should ensure that, to receive correct echo with certain probability, under two-dimensional Cartesian coordinate system, if measure to
Amount be bidimensional, it is determined that tracking gate area be Av=π γ | S (k) |1/2, tracking gate γ=16, then judge candidate measurement
Whether point meets:
Wherein, zkFor the measurement vector of k moment measuring point,For the mesh predicted at the k-1 moment the k moment
Target predicted vector, parameter γ is by the χ under the tracking gate rule2Distribution table obtains.
Bo Mennei will be fallen into meet candidate measuring point of the measuring point as the k moment of above formula condition.
Step 5, when the number of the candidate measuring point at k moment be 0 when, then the state estimation of the k moment target in step 3 to
AmountWith evaluated error covariance matrix Pk/k-1For course extrapolation.
When the number of the candidate measuring point at k moment is 1, then the measurement vector of the candidate measuring point at k moment is used for track
It updates.
When the number of the candidate measuring point at k moment is greater than 1, optimized using targetpath of the entropy assessment to the k moment
It updates.
It is described that update is optimized using targetpath of the entropy assessment to the k moment, include following sub-step:
Sub-step 5.1, according to the l of a candidate measuring point of the jth (j=1,2 ..., m, m > 1) at k moment (l=1,
2 ..., n, n >=1) a measurement index ajlWith the predicted vector of k moment targetCalculate k moment j-th candidates measuring point
First of measurement index ajlWith first of measurement index a of the prediction measuring point i at tracking gate centerilBetween calculating error
It is worth Δjl:
Δjl=| ajl-ail|；
Specific step is as follows:
As shown in Figure 1, when there is no measuring point to fall into related Bo Mennei, then the state of the k moment target calculated in step 3
Estimate vectorWith evaluated error covariance matrix Pk/k-1It is directly used in course extrapolation.
If the measuring point for falling into related Bo Mennei only has 1, which can be directly used in track update.
If there is more than one measuring point to fall in the related Bo Mennei of tracked target, to the candidate measuring point at k moment
Information matrix A is formed according to respective measurement index l (l=1,2 ..., n, n >=1), if setting radar tracking system has m candidate
Measuring point, each candidate's measuring point have n measurement index, then information matrix A are as follows:
Wherein, each n measurements index for being classified as a candidate measuring point in A, m candidate's measurement of each behavior in A
The same measurement index of point, ajlFor first of measurement index of j-th candidates measuring point.
The measurement index prediction corresponding with tracking gate center of candidate's measuring point each in information matrix A is measured
The matrix of absolute value of the difference composition between the corresponding measurement index of point is denoted as information error matrix ΔA:
Wherein, ΔAIn each n measurement index for being classified as a candidate measuring point and corresponding measurement index predicted value
Between absolute value of the difference；ΔjlFor first of measurement index a of j-th candidates measuring pointjlWith first of amount of prediction measuring point i
Survey index ailBetween absolute value of the difference, i.e. Δjl=| ajl-ail|。
Sub-step 5.2 is measured according to k moment j-th candidates measuring point is corresponding with the prediction measuring point i at tracking gate center
Error amount Δ between indexjlWith the quantity m of k moment candidate measuring point, the uncertainty of first of k moment measurement index is obtained
Sl；The uncertainty S of index is measured to first of the k momentlIt is normalized, obtains the weight of first of k moment measurement index
Factor alphal。
Sub-step 5.2 includes following sub-step:
Sub-step 5.2.1, according to first of measurement index a of k moment j-th candidates measuring pointjlWith prediction measuring point i's
First of measurement index ailBetween error amount Δjl, first of measurement index of k moment j-th candidates measuring point is calculated in m
Shared ratio p in a candidate's measuring pointjl:
Sub-step 5.2.2, according to first of measurement index of k moment j-th candidates measuring point in m candidate measuring point
Shared ratio pjl, calculate the comentropy H of first of k moment measurement indexl:
Sub-step 5.2.3 calculates first of measurement index of k moment according to the quantity m of k moment j-th candidates measuring point
Maximum informational entropy Hlmax。
Specifically, the comentropy of a measurement index is bigger, and the error amount represented between the measurement and prediction measurement more connects
Closely；Conversely, represent the measurement and prediction measure between error amount further away from.Work as pjlWhen=1/m, i.e., each candidate measuring point exists
When the probability of proportion is the same under the measurement index, the maximum informational entropy for calculating first of k moment measurement index is Hlmax=
log2m。
Sub-step 5.2.4 measures the comentropy H of index according to first of the k momentlMost with first of measurement index of k moment
Big comentropy Hlmax, calculate the uncertainty S of first of k moment measurement indexl:
Sub-step 5.2.5 measures the uncertainty S of index using first of the k momentl, calculate first of the measurement of k moment and refer to
Target weight coefficient αl:
By being distributed the method being embodied to measurement information using comentropy, if the comentropy for measuring index is smaller,
The information content that the measurement index provides is bigger, acts on played in overall merit then bigger.
Sub-step 5.3 measures the weight coefficient α of index according to first of the k momentl, using statistical distance relation maxim to k
Statistical distance between moment prediction measuring point i and candidate measuring point carries out tax power processing, obtains the tax power statistical distance at k moment
IEWdIj, k；Statistical distance IEWd is weighed in the tax for choosing the k momentIj, kCorresponding candidate measuring point is preferred as the k moment when minimum
Target measuring point optimizes update to the targetpath at k moment.
Step 5.3 includes following sub-step:
Sub-step 5.3.1, according to the residual error e of k moment Kalman filterIj, k, obtaining the k moment predicts measuring point i and jth
Statistical distance d between a candidate's measuring pointIj, k；Include following sub-step:
Sub-step 5.3.1.1, setting Bo Mennei have m candidate measuring point, and each candidate's measuring point includes that n measurement refers to
It marks, then the Kalman filter residual error e at k momentIj, kAre as follows:
Wherein, zJ, kFor the measurement vector of k moment j-th candidates measuring point,It is pre- to the k moment at the k-1 moment
Measure the state estimation vector of measuring point i.
Sub-step 5.3.1.2, according to the residual error e of k moment Kalman filterIj, k, calculate the prediction measuring point at k moment
The statistical distance d of i and j-th candidates measuring pointIj, kAre as follows:
eIj, k=[Δ1..., Δl..., Δn]T；
Wherein, SIj, kThe new breath covariance matrix of measuring point i Yu j-th candidates measuring point, Δ are predicted for the k moment1For jth
The 1st of a candidate's measuring point measures the absolute value of the difference between index and the corresponding measurement index for predicting measuring point i, ΔlFor
The l (l=1,2 ..., n) of j-th candidates measuring point is a to be measured between index and the corresponding measurement index for predicting measuring point i
Absolute value of the difference, ΔnBetween n-th of measurement index of j-th candidates measuring point and the corresponding measurement index for predicting measuring point i
Absolute value of the differenceThe new breath variance of index is measured for first of the k moment.
It is identical that index weighting weight is respectively measured in the above statistical distance calculation formula, in statistical distance, i.e. weight coefficient phase
Together, the various information for measuring index are not made full use of, are occurred so as to cause nearest-neighbor algorithm under statistical distance biggish
Deviation.
5.3.2, the weight coefficient α of index is measured using first of the k momentlMeasuring point i and j-th candidates are predicted to the k moment
Statistical distance between measuring point is weighted processing, obtains entitled statistical distance IEWdIj, k:
Wherein, αlFor a weight coefficient for measuring index of k moment l (l=1,2 ..., n), ΔlFor j-th candidates measurement
Absolute value of the difference between the measurement index l (l=1,2 ..., n) of point measurement index corresponding with prediction measuring point,When for k
Carve the new breath variance of first of measurement index.
The entropy that some measures index is smaller, illustrates that the mutation level of the measurement index is bigger.Radar tracking system can be with
More information content is obtained from the measurement index, in general all measurement indexs, the small measurement index of entropy is in systems
Effect it is also bigger, so biggish weight should be assigned.Conversely, the entropy for measuring index is bigger, illustrate the measurement index
Mutation level it is smaller, system can obtain less information content, it should assign lesser weight.
5.3.3, according to k moment entitled statistical distance IEWdIj, k, choose the tax power statistical distance IEWd at k momentIj, kMost
Selected objective target measuring point of the candidate measuring point as the k moment corresponding to hour, optimizes more the targetpath at k moment
Newly.
IEWdIj, kFor the statistical distance based on information entropy weight, for calculating between candidate measuring point and prediction measuring point
Distance takes IEWdIj, kSelected objective target measuring point of the corresponding candidate measuring point as k moment when minimum, to the target at k moment
Track optimizes update.
By that can further reduce tracking and miss using the method for measuring indication information entropy and entropy assessment being introduced statistical distance
Difference.
In the present invention, according to the measurement of the smallest optimization aim measuring point of the obtained statistical distance based on information entropy weight to
Measure zk', the state estimation vector of the k moment target after calculating optimizationWith evaluated error covariance matrix Pk。
Pk=Pk/k-1-KkH Pk/k-1
According to the state estimation vector of the k moment target after optimizationWith evaluated error covariance matrix Pk, to the k+1 moment
The state estimation vector of targetWith evaluated error covariance matrix Pk+1/kIt is predicted, obtains the state of k+1 moment target
Predicted vectorNew breath covariance matrix Sk+1With kalman gain Kk+1, then, step 3 is repeated to step 5, to the k+1 moment
Targetpath optimize update, and recursion goes out the state estimation vector of k+1 moment targetWith evaluated error covariance square
Battle array Pk+1, carry out repeatedly, obtain the optimization track information of the target.
Emulation experiment:
It sets and uses two kinds of target tracking algorisms, nearest-neighbor algorithm (NN) and inventive algorithm in simulation process altogether
It (IEWNN) is all that target following is carried out to the single goal under clutter environment.It is assumed that the target following scene under a kind of clutter environment,
Emulation under two-dimensional plane coordinate system, if the measurement vector of measuring point be bidimensional, it is determined that tracking gate area be Av=π γ | S
(k)|1/2, wherein tracking gate is set as γ=16, SkNewly to cease covariance matrix, false measure is in unit area
0.0089, the false number that measures is 50 in per time instance.Target makees uniform rectilinear (CV) movement, system state equation such as step 1 institute
Show, target initial state vector is x0=[200 0 10000-35]T, the measurement equation of radar tracking system as shown in step 1,
Measure noise WkFor variance r=200m2Zero mean Gaussian white noise, initial covariance matrix be R0, and R11=R22=200m2,
R12=R21=0m2, sampling interval t=1s, false measurement is not added by preceding 10s in target motion process, adds from 10s to 200s
Enter false measurement.
Radar is carried out continuously 200 scanning, obtains tracking based on nearest-neighbor algorithm with being associated with based on the method for the present invention
The target following track plot arrived, as shown in figure 3, from figure 3, it can be seen that the track of the method for the present invention is closer to true feelings
Condition illustrates that the method for the present invention reduces data correlation error.
Fig. 4 is the target following pair of nearest-neighbor algorithm (NN) and the nearest-neighbor algorithm (IEWNN) based on information entropy weight
Than figure, from Fig. 4 (a), 4 (b), 4 (c), 4 (d) as can be seen that being compared to nearest-neighbor algorithm, X, Y after the method for the present invention tracking
The estimated value convergence rate of the distance in direction and speed faster, convergence effect it is more excellent, actual tracking situation is closer to true value.Figure
5 is using the comparisons of the target following error of nearest-neighbor algorithm (NN) and the nearest-neighbor algorithm (IEWNN) based on information entropy weight
Figure, as can be seen that being compared to nearest-neighbor algorithm from Fig. 5 (a), 5 (b), 5 (c), 5 (d), X, Y after the method for the present invention tracking
The error convergence speed of the distance in direction and speed faster, closer to X, the actual distance of Y-direction and speed.Fig. 6 is to use
Nearest-neighbor algorithm (NN) and nearest-neighbor algorithm (IEWNN) based on information entropy weight carry out 100 Monte Carlo Experiments respectively
Root-mean-square error comparison diagram, as can be seen that be compared to nearest-neighbor algorithm from Fig. 6 (a), 6 (b), 6 (c), 6 (d), this
Each moment X of inventive method, Y-direction distance and direction speed root-mean-square error are superior to NN algorithm.
In summary it is found that the resulting radar tracking result of the method for the present invention is closer with truth, illustrate the present invention
Method passes through the comentropy of measurement information known to analytical calculation, and each weight coefficient for measuring index is determined using entropy assessment,
Tax power is carried out to statistical distance relation maxim, statistical distance criterion is solved since the participation of all properties equalization calculates, does not have
The shortcomings that significance level of prominent attribute, as a result influence vulnerable to single attribute.Have compared to other innovatory algorithms lower
Improvement cost and better applicability, and it is more amenable for use with the transformation and upgrade of relevant device.The present invention theoretically not only limits to
Lower nearest-neighbor algorithm improvement optimization is measured in two dimension, is applied equally to nearest-neighbor algorithm under multidimensional measurement information, and
As measurement information gets over diversification, the information excavated is also abundanter, will be more ideal to algorithm improvement effect.
First of measurement index in the embodiment of the present invention inscribes Trajectory Prediction measuring point corresponding two-dimentional right angle when being corresponding
The information such as X, the distance of Y-direction, speed, azimuth, amplitude under coordinate system.Ture in description of the invention attached drawing is target
Truth.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (10)
1. the targetpath optimization method based on information entropy weight and nearest-neighbor data correlation, which is characterized in that including following step
It is rapid:
Step 1, the state equation of radar tracking system is set as Xk+1=FXkMeasurement equation with radar tracking system is Zk=
HXk+Wk；
The measuring point that the first two moment radar tracking system after setting track initiation obtains is unique substantial amount measuring point；Using
The measurement vector Z of the measuring point at the first two moment after the track initiation that radar tracking system obtains0, Z1, calculate the initial of target
State vector
Wherein, Xk+1For the state vector of k+1 moment radar tracking system, XkFor the state vector of k moment radar tracking system, Wk
For the measurement noise sequence at k moment；F is the state-transition matrix of radar tracking system, and H is the measurement square of radar tracking system
Battle array；x0For the position of target X-direction under initial time two-dimensional Cartesian coordinate system, y0For mesh under initial time two-dimensional Cartesian coordinate system
The position of Y-direction is marked,For the speed of target X-direction under initial time two-dimensional Cartesian coordinate system,It is straight for initial time two dimension
The speed of target Y-direction, [] under angular coordinate systemTFor the transposition of matrix；
Step 2, according to the measurement vector Z of the measuring point at the first two moment after the track initiation of radar tracking system acquisition0, Z1, lead to
It crosses two o'clock calculus of finite differences to initialize Kalman filter, calculates the Initial state estimation vector of Kalman filterWith
Initial estimation error co-variance matrix P0；
Step 3, according to the Initial state estimation vector of Kalman filterWith initial estimation error co-variance matrix P0, use
Iterative method calculates k (k=1,2...n, n ∈ N+) moment target state estimation vectorThe evaluated error of k moment target
Covariance matrix Pk/k-1, k moment target predicted vectorThe new breath covariance matrix S of k moment targetkWith k moment target
Kalman gain Kk；
Step 4, the measurement point set W that the radar scanning of k moment obtains is obtainedk, by the predicted vector of the k moment targetPosition
It sets as tracking gate center, tracking gate threshold value is chosen, to the measurement point set W at k momentkPreliminary screening is carried out, k is obtained
The candidate measuring point at moment；
Step 5, when the number of the candidate measuring point at k moment is 0, then the state estimation vector of the k moment targetAnd institute
State the evaluated error covariance matrix P of k moment targetk/k-1For course extrapolation；
When the number of the candidate measuring point at k moment is 1, then the measurement vector of the candidate measuring point at the k moment is used for track
It updates；
When the number of the candidate measuring point at k moment is greater than 1, update is optimized using targetpath of the entropy assessment to the k moment.
2. the targetpath optimization method according to claim 1 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in step 1, sets under two dimensional motion, the state-transition matrix F of the radar tracking system, the radar tracking
The measurement vector Z of the measurement matrix H of system, the k moment targetkWith the measuring point at the first two moment after the track initiation
Measure vector Z0, Z1Expression formula be respectively as follows:
Zk=[ZX, k ZY, k]T, Z0=[ZX, 0 ZY, 0]T, Z1=[ZX, 1 ZY, 1]T；
Wherein, []TFor the transposition of matrix, t is the sampling interval；ZX, kFor the k moment under two-dimensional Cartesian coordinate system target X-direction
Distance；ZY, kFor the distance of k moment target Y-direction under two-dimensional Cartesian coordinate system；ZX, 1For target the 1st moment obtain two
Tie up X-direction position measuring value under rectangular coordinate system, ZX, 0For target under the two-dimensional Cartesian coordinate system that the 0th moment obtains X-direction
Position measuring value, ZY, 1For target under the two-dimensional Cartesian coordinate system that the 1st moment obtains Y-direction position measuring value, ZY, 0For mesh
It is marked on Y-direction position measuring value under the two-dimensional Cartesian coordinate system of the 0th moment acquisition.
3. the targetpath optimization method according to claim 2 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in step 2, the calculation formula that Kalman filter is initialized by two o'clock calculus of finite differences are as follows:
Wherein, r is the measurement noise sequence W of initial time0Variance,For Kalman filter under two-dimensional Cartesian coordinate system
X-direction position estimation value when original state,For X-direction when Kalman filter original state under two-dimensional Cartesian coordinate system
Velocity estimation value,For Y-direction position estimation value when Kalman filter original state under two-dimensional Cartesian coordinate system,
For Y-direction velocity estimation value when Kalman filter original state under two-dimensional Cartesian coordinate system.
4. the targetpath optimization method according to claim 3 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in step 3, formula used by the iterative method are as follows:
Pk/k-1=FPk-1FT
Sk=HPk/k-1HT+Rk
Wherein,For the state estimation vector of k-1 moment target, Pk-1For the evaluated error covariance matrix of k-1 moment target,
RkNoise covariance matrix is measured for the k moment,TFor the transposition of matrix,-1For inverse of a matrix.
5. the targetpath optimization method according to claim 1 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in step 4, the preliminary screening includes following sub-step:
Sub-step 4.1 is set under two-dimensional Cartesian coordinate system, and tracking gate is chosen according to elliptical wave door rule, the measurement of target to
Amount is two dimension, determines that tracking gate area is Av=π γ | Sk|1/2；
Sub-step 4.2 judges whether candidate measuring point meets tracking gate threshold condition according to tracking gate area:
Using the measuring point for the condition that meets as the candidate measuring point at k moment.
Wherein, zkFor the measurement vector of k moment measuring point,For the pre- of the target predicted at the k-1 moment to the k moment
Direction finding amount；γ is parameter, by the χ under elliptical wave door rule2Distribution table obtains.
6. the targetpath optimization method according to claim 1 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in step 5, described optimized using targetpath of the entropy assessment to the k moment is updated comprising following sub-step:
Sub-step 5.1, according to l (l=1,2 ..., n, n of a candidate measuring point of the jth (j=1,2 ..., m, m > 1) at k moment
>=1) a measurement index ajlWith the predicted vector of k moment targetCalculate first of amount of k moment j-th candidates measuring point
Survey index ajlWith first of measurement index a of the prediction measuring point i at tracking gate centerilBetween error amount Δjl:
Δjl=| ajl-ail|；
Step 5.2, according to k moment j-th candidates measuring point measurement index corresponding with the prediction measuring point i at tracking gate center it
Between error amount ΔjlWith the quantity m of k moment candidate measuring point, the uncertainty S of first of k moment measurement index is obtainedl；To k
The uncertainty S of first of moment measurement indexlIt is normalized, obtains the weight coefficient of first of k moment measurement index
αl；
Step 5.3, the weight coefficient α of index is measured according to first of the k momentl, the k moment is predicted using statistical distance relation maxim
Statistical distance between measuring point i and candidate measuring point carries out tax power processing, obtains the tax power statistical distance IEWd at k momentIj, k；
Statistical distance IEWd is weighed in the tax for choosing the k momentIj, kCorresponding candidate's measuring point is measured as the selected objective target at k moment when minimum
Point optimizes update to the targetpath at k moment.
7. the targetpath optimization method according to claim 6 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, sub-step 5.2 includes following sub-step:
Sub-step 5.2.1, according to first of measurement index a of k moment j-th candidates measuring pointjlWith first of prediction measuring point i
Measure index ailBetween error amount Δjl, first of measurement index of k moment j-th candidates measuring point is calculated in m candidate
Shared ratio p in measuring pointjl:
Sub-step 5.2.2, it is shared in m candidate measuring point according to first of measurement index of k moment j-th candidates measuring point
Ratio pjl, calculate the comentropy H of first of k moment measurement indexl:
Sub-step 5.2.3 calculates first of measurement index of k moment most according to the quantity m of k moment j-th candidates measuring point
Big comentropy Hlmax:
Hlmax=log2m；
Sub-step 5.2.4 measures the comentropy H of index according to first of the k momentlThe maximum information of index is measured with first of the k moment
Entropy Hlmax, calculate the uncertainty S of first of k moment measurement indexl:
Sub-step 5.2.5 measures the uncertainty S of index using first of the k momentl, calculate first of k moment measurement index
Weight coefficient αl:
8. the targetpath optimization method according to claim 7 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, step 5.3 includes following sub-step:
Sub-step 5.3.1, according to the residual error e of k moment Kalman filterIj, k, obtain the k moment predict measuring point i and j-th time
Select the statistical distance d between measuring pointIj, k；
Sub-step 5.3.2 measures the weight coefficient α of index using first of the k momentlMeasuring point i and j-th of time are predicted to the k moment
It selects the statistical distance between measuring point to be weighted processing, obtains the tax power statistical distance IEWd at k momentIj, k；
Sub-step 5.3.3 weighs statistical distance IEWd according to the tax at k momentIj, k, choose the tax power statistical distance IEWd at k momentIj, k
Selected objective target measuring point of the corresponding candidate's measuring point as the k moment, optimizes more the targetpath at k moment when minimum
Newly.
9. the targetpath optimization method according to claim 8 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, sub-step 5.3.1 includes following sub-step:
Sub-step 5.3.1.1, setting in tracking gate has m candidate measuring point, and each candidate's measuring point includes that n measurement refers to
It marks, then the Kalman filter residual error e at k momentIj, kAre as follows:
Wherein, zJ, kFor the measurement vector of k moment j-th candidates measuring point,To be measured at the k-1 moment to the prediction at k moment
The state estimation vector of point i；
Sub-step 5.3.1.2, according to the residual error e of k moment Kalman filterIj, k, calculate the prediction measuring point i and at k moment
The statistical distance d of j candidate measuring pointIj, kAre as follows:
eIj, k=[Δ1..., Δl..., Δn]T；
Wherein, SIj, kThe new breath covariance matrix of measuring point i Yu j-th candidates measuring point, Δ are predicted for the k momentlFor k moment jth
The l (l=1,2 ..., n, n >=1) of a candidate's measuring point is a to be measured between index and the corresponding measurement index for predicting measuring point i
Absolute value of the difference,The new breath variance of index is measured for first of the k moment.
10. the targetpath optimization method according to claim 9 based on information entropy weight and nearest-neighbor data correlation,
It is characterized in that, in sub-step 5.3.2, the weighting processing are as follows:
Wherein, IEWdIj, kStatistical distance, α are weighed for the tax at k momentlThe weight coefficient of index is measured for first of the k moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910248309.5A CN109901153B (en) | 2019-03-29 | 2019-03-29 | Target track optimization method based on information entropy weight and nearest neighbor data association |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910248309.5A CN109901153B (en) | 2019-03-29 | 2019-03-29 | Target track optimization method based on information entropy weight and nearest neighbor data association |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109901153A true CN109901153A (en) | 2019-06-18 |
| CN109901153B CN109901153B (en) | 2022-11-01 |
Family
ID=66954041
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910248309.5A Active CN109901153B (en) | 2019-03-29 | 2019-03-29 | Target track optimization method based on information entropy weight and nearest neighbor data association |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109901153B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110780289A (en) * | 2019-10-23 | 2020-02-11 | 北京信息科技大学 | Multi-target vehicle tracking method and device based on scene radar |
| CN110824467A (en) * | 2019-11-15 | 2020-02-21 | 中山大学 | Multi-target tracking data association method and system |
| CN110990756A (en) * | 2019-12-03 | 2020-04-10 | 中国电子科技集团公司第二十八研究所 | Aerial target track quality assessment method based on extension type layer-by-layer analysis |
| CN111007495A (en) * | 2019-12-10 | 2020-04-14 | 西安电子科技大学 | Target track optimization method based on double-fusion maximum entropy fuzzy clustering JPDA |
| CN111289965A (en) * | 2019-12-04 | 2020-06-16 | 南京长峰航天电子科技有限公司 | Multi-target radar rapid tracking method and system |
| CN111398948A (en) * | 2020-04-08 | 2020-07-10 | 成都汇蓉国科微系统技术有限公司 | Maneuvering small target track association method under strong clutter background |
| CN111983926A (en) * | 2020-08-31 | 2020-11-24 | 郑州轻工业大学 | Maximum co-entropy extended ellipsoid collective filtering method |
| CN111983927A (en) * | 2020-08-31 | 2020-11-24 | 郑州轻工业大学 | Novel maximum entropy ellipsoid collective filtering method |
| CN112098993A (en) * | 2020-09-16 | 2020-12-18 | 中国北方工业有限公司 | Multi-target tracking data association method and system |
| CN112526521A (en) * | 2020-11-25 | 2021-03-19 | 湖北工业大学 | Multi-target tracking method for automobile millimeter wave anti-collision radar |
| CN113408392A (en) * | 2021-06-15 | 2021-09-17 | 西安电子科技大学 | Flight path completion method based on Kalman filtering and neural network |
| CN114139995A (en) * | 2021-12-06 | 2022-03-04 | 中国科学技术信息研究所 | Test area monitoring and evaluating method and device, electronic equipment and storage medium |
| CN114459484A (en) * | 2022-01-28 | 2022-05-10 | 江苏涵润汽车电子有限公司 | Track initialization method and device |
| RU2776078C1 (en) * | 2021-07-02 | 2022-07-13 | Российская Федерация, От Имени Которой Выступает Министерство Промышленности И Торговли Российской Федерации | Method for determining the coordinates and motion parameters of radio emission sources based on measurements of their carrier frequency |
| CN116228018A (en) * | 2023-03-02 | 2023-06-06 | 中国人民解放军63921部队 | Early warning equipment efficiency evaluation method based on non-target theoretical truth value |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6801152B1 (en) * | 2003-04-21 | 2004-10-05 | Northrop Grumman Corporation | Method for determining the optimum observer heading change in bearings-only passive emitter tracking |
| US20140139367A1 (en) * | 2012-11-22 | 2014-05-22 | Agency For Defense Development | Method for producing at least information for track fusion and association for radar target tracking, and storage medium thereof |
| CN104155650A (en) * | 2014-08-14 | 2014-11-19 | 电子科技大学 | Object tracking method based on trace point quality evaluation by entropy weight method |
| CN108802692A (en) * | 2018-05-25 | 2018-11-13 | 哈尔滨工程大学 | A kind of method for tracking target based on maximum cross-correlation entropy volume particle filter |
| CN108983215A (en) * | 2018-05-25 | 2018-12-11 | 哈尔滨工程大学 | A kind of method for tracking target based on maximum cross-correlation entropy adaptively without mark particle filter |
| CN109002835A (en) * | 2018-06-19 | 2018-12-14 | 西安电子科技大学 | A kind of particle filter data correlation method based on maximum entropy fuzzy clustering |
-
2019
- 2019-03-29 CN CN201910248309.5A patent/CN109901153B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6801152B1 (en) * | 2003-04-21 | 2004-10-05 | Northrop Grumman Corporation | Method for determining the optimum observer heading change in bearings-only passive emitter tracking |
| US20140139367A1 (en) * | 2012-11-22 | 2014-05-22 | Agency For Defense Development | Method for producing at least information for track fusion and association for radar target tracking, and storage medium thereof |
| CN104155650A (en) * | 2014-08-14 | 2014-11-19 | 电子科技大学 | Object tracking method based on trace point quality evaluation by entropy weight method |
| CN108802692A (en) * | 2018-05-25 | 2018-11-13 | 哈尔滨工程大学 | A kind of method for tracking target based on maximum cross-correlation entropy volume particle filter |
| CN108983215A (en) * | 2018-05-25 | 2018-12-11 | 哈尔滨工程大学 | A kind of method for tracking target based on maximum cross-correlation entropy adaptively without mark particle filter |
| CN109002835A (en) * | 2018-06-19 | 2018-12-14 | 西安电子科技大学 | A kind of particle filter data correlation method based on maximum entropy fuzzy clustering |
Non-Patent Citations (2)
| Title |
|---|
| CHUNXIA LI: "A Novel Method to Analyze Accuracy of Target Measurement Based on Entropy", 《2017 4TH INTERNATIONAL CONFERENCE ON INFORMATION SCIENCE AND CONTROL ENGINEERING》 * |
| 惠黎明: "基于多传感器探测精度的正确融合航迹判断", 《电子科技》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110780289B (en) * | 2019-10-23 | 2021-07-30 | 北京信息科技大学 | Multi-target vehicle tracking method and device based on scene radar |
| CN110780289A (en) * | 2019-10-23 | 2020-02-11 | 北京信息科技大学 | Multi-target vehicle tracking method and device based on scene radar |
| CN110824467A (en) * | 2019-11-15 | 2020-02-21 | 中山大学 | Multi-target tracking data association method and system |
| CN110824467B (en) * | 2019-11-15 | 2021-11-19 | 中山大学 | Multi-target tracking data association method and system |
| CN110990756A (en) * | 2019-12-03 | 2020-04-10 | 中国电子科技集团公司第二十八研究所 | Aerial target track quality assessment method based on extension type layer-by-layer analysis |
| CN111289965A (en) * | 2019-12-04 | 2020-06-16 | 南京长峰航天电子科技有限公司 | Multi-target radar rapid tracking method and system |
| CN111007495A (en) * | 2019-12-10 | 2020-04-14 | 西安电子科技大学 | Target track optimization method based on double-fusion maximum entropy fuzzy clustering JPDA |
| CN111398948A (en) * | 2020-04-08 | 2020-07-10 | 成都汇蓉国科微系统技术有限公司 | Maneuvering small target track association method under strong clutter background |
| CN111398948B (en) * | 2020-04-08 | 2021-12-10 | 成都汇蓉国科微系统技术有限公司 | Maneuvering small target track association method under strong clutter background |
| CN111983927A (en) * | 2020-08-31 | 2020-11-24 | 郑州轻工业大学 | Novel maximum entropy ellipsoid collective filtering method |
| CN111983926A (en) * | 2020-08-31 | 2020-11-24 | 郑州轻工业大学 | Maximum co-entropy extended ellipsoid collective filtering method |
| CN111983927B (en) * | 2020-08-31 | 2022-04-12 | 郑州轻工业大学 | Ellipsoid collective filtering method for maximum covariance MCC (MCC) criterion |
| CN112098993A (en) * | 2020-09-16 | 2020-12-18 | 中国北方工业有限公司 | Multi-target tracking data association method and system |
| CN112526521A (en) * | 2020-11-25 | 2021-03-19 | 湖北工业大学 | Multi-target tracking method for automobile millimeter wave anti-collision radar |
| CN112526521B (en) * | 2020-11-25 | 2022-08-19 | 湖北工业大学 | Multi-target tracking method for automobile millimeter wave anti-collision radar |
| CN113408392A (en) * | 2021-06-15 | 2021-09-17 | 西安电子科技大学 | Flight path completion method based on Kalman filtering and neural network |
| RU2776078C1 (en) * | 2021-07-02 | 2022-07-13 | Российская Федерация, От Имени Которой Выступает Министерство Промышленности И Торговли Российской Федерации | Method for determining the coordinates and motion parameters of radio emission sources based on measurements of their carrier frequency |
| CN114139995B (en) * | 2021-12-06 | 2022-07-26 | 中国科学技术信息研究所 | Test area monitoring and evaluating method and device, electronic equipment and storage medium |
| CN114139995A (en) * | 2021-12-06 | 2022-03-04 | 中国科学技术信息研究所 | Test area monitoring and evaluating method and device, electronic equipment and storage medium |
| CN114459484A (en) * | 2022-01-28 | 2022-05-10 | 江苏涵润汽车电子有限公司 | Track initialization method and device |
| CN114459484B (en) * | 2022-01-28 | 2024-05-28 | 江苏涵润汽车电子有限公司 | Track initialization method and device |
| CN116228018A (en) * | 2023-03-02 | 2023-06-06 | 中国人民解放军63921部队 | Early warning equipment efficiency evaluation method based on non-target theoretical truth value |
| CN116228018B (en) * | 2023-03-02 | 2024-03-01 | 中国人民解放军63921部队 | Early warning equipment efficiency evaluation method based on non-target theoretical truth value |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109901153B (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109901153A (en) | Targetpath optimization method based on information entropy weight and nearest-neighbor data correlation | |
| CN111007495B (en) | Target track optimization method based on double fusion maximum entropy fuzzy clustering JPDA | |
| CN109002835B (en) | Particle filter data association method based on maximum entropy fuzzy clustering | |
| CN109633589A (en) | The Multi-target Data Associations assumed are optimized based on multi-model more in target following | |
| CN109508000A (en) | Isomery multi-sensor multi-target tracking method | |
| CN108303692A (en) | A kind of multi-object tracking method of solution doppler ambiguity | |
| CN101614817A (en) | A kind of multi-object tracking method based on ground moving target indication radar system | |
| CN109752710B (en) | Rapid target angle estimation method based on sparse Bayesian learning | |
| CN109946694A (en) | Circumference SAR multi-object tracking method based on stochastic finite collection | |
| CN106021697B (en) | A kind of rapid phase-control battle array radar Time-energy resource joint management method | |
| CN108152789A (en) | Utilize the passive track-corelation data correlation and localization method of RSS information | |
| CN109539884A (en) | A kind of Vector Target-missing Quantity method for parameter estimation based on GA | |
| Lan et al. | A survey on joint tracking using expectation–maximization based techniques | |
| CN111460597A (en) | Radar station distribution method based on improved multi-target particle swarm optimization algorithm | |
| CN111711432B (en) | Target tracking algorithm based on UKF and PF hybrid filtering | |
| CN110849372A (en) | Underwater multi-target track association method based on EM clustering | |
| CN111259332B (en) | Fuzzy data association method and multi-target tracking method in clutter environment | |
| Anderson et al. | Track association for over-the-horizon radar with a statistical ionospheric model | |
| CN110297221A (en) | A kind of data correlation method based on gauss hybrid models | |
| CN111274529B (en) | Robust Gao Sini Weisal PHD multi-expansion target tracking algorithm | |
| KR101045348B1 (en) | Target tracking method using doppler velocity in rectangular coordinate system | |
| Rawat et al. | Design of a performance evaluation methodology for data-fusion-based multiple target tracking systems | |
| CN110488276A (en) | The optimal resource allocation method based on demand of isomery radar fence towards multiple target tracking task | |
| Lee et al. | Robust measurement validation for radar target tracking using prior information | |
| Huang et al. | Joint integrated track splitting for multi-path multi-target tracking using OTHR detections |
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 |