CN112415503A - Multi-target particle filter pre-detection tracking method based on target re-tracking - Google Patents

Multi-target particle filter pre-detection tracking method based on target re-tracking Download PDF

Info

Publication number
CN112415503A
CN112415503A CN202011102472.XA CN202011102472A CN112415503A CN 112415503 A CN112415503 A CN 112415503A CN 202011102472 A CN202011102472 A CN 202011102472A CN 112415503 A CN112415503 A CN 112415503A
Authority
CN
China
Prior art keywords
target
particle
tracking
radar
calculating
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
Application number
CN202011102472.XA
Other languages
Chinese (zh)
Other versions
CN112415503B (en
Inventor
陈霄
潘凯
刘光宏
韩阔业
薛安克
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN202011102472.XA priority Critical patent/CN112415503B/en
Publication of CN112415503A publication Critical patent/CN112415503A/en
Application granted granted Critical
Publication of CN112415503B publication Critical patent/CN112415503B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/66—Radar-tracking systems; Analogous systems
    • G01S13/72—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar
    • G01S13/723—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar by using numerical data
    • G01S13/726—Multiple target tracking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a multi-target particle filter pre-detection tracking method based on target retracing, which is characterized in that in a tracking link, for targets which are tracked and pb of which is greater than a given threshold, tracking particle groups are sorted in a reverse order according to the weight value of particles, and the state mean value of the first N/50 particles is calculated. And generating new particles by taking the state mean value as a central point, replacing state information of the last 0.98 × N particles of the tracking particle swarm, setting the existing values of the particle swarm to be 1, reserving high-quality particles, updating the particle swarm, improving the particle utilization rate, enabling the tracking point trace to be close to a real target quickly, and improving the tracking precision of the target. In addition, the false trace point is enabled to be more quickly close to the real target, and the length len of the trace track of the target is calculatediIf the distance between the target and the first track point is 3, the initial state information of the target is inaccurate, the false track is deleted, and the false alarm rate of the target are reducedA tracking error.

Description

Multi-target particle filter pre-detection tracking method based on target re-tracking
Technical Field
The invention belongs to the technical field of tracking before radar detection, relates to the technical field of tracking before multi-radar multi-target particle filter detection, and particularly relates to a multi-target particle filter tracking-before-detection method based on target re-tracking
Background
The multi-radar multi-target particle filter pre-detection tracking algorithm is a method for detecting and tracking a plurality of weak targets by using a plurality of radars, and a double-layer particle filter structure, namely a target tracking layer and a target detection layer, is usually adopted. The target detection layer is responsible for detecting and finding new targets, the target tracking layer is responsible for carrying out independent tracking estimation on each target found at present, and each target has an independent tracking particle swarm.
When the target tracking layer carries out tracking filtering on the found targets, the tracking precision of the newly found targets is generally lower and the newly found targets are far away from the real tracks of the targets, so when the distances of a plurality of targets are close, if the tracking point tracks of the newly found targets are close to the real tracks of another target, because the tracking threshold of the target tracking layer is large, the tracking particle swarm of the newly found targets is easy to deviate from the original target tracks, gradually approaches to the other target tracks, and even is combined with the other target tracks, and the target missing detection problem is finally caused.
Disclosure of Invention
The invention provides a multi-target particle filter pre-detection tracking method based on target re-tracking, which considers the problem of target omission caused by high target false alarm rate and high tracking error in the early tracking period in the tracking of a plurality of targets.
The method comprises the following specific steps:
step 1, initializing parameters: radar scanning period T, total observation frame number K, number of particles in particle swarm N, radar number R, frame length len of target tracking trackiDistance, Doppler and azimuth space cell distance are Dr,Dd,DbError threshold Error;
step 2, reading the k frame measurement of multiple radars
Figure BDA0002725849740000011
Wherein the content of the first and second substances,
Figure BDA0002725849740000012
the measurement in a measurement unit (m, n, p) of the echo data of the kth frame of the r-th radar is shown, wherein m, n and p respectively show a distance unit, a Doppler unit and a direction unit;
step 3, setting the tracking target set Taxe at the moment k-1 as f1,k-1，,f2,k-1…fTm,k-1Tracking the targets in the data, wherein Tm is the number of the targets in the tracking set, and each target fi,k-1All have a tracking particle swarm Pi,k-1＝{pi,1,k-1,pi,2,k-1…pi,N,k-1In which p isi,j,k-1Representing the jth particle at the time of the ith target k-1.
Step 3.1, changing i to 1;
step 3.2, tracking particle swarm P of ith targeti,k-1＝{pi,1,k-1,pi,2,k-1…pi,N,k-1Performing state transition on each particle in the target particle swarm P to obtain a target tracking particle swarm Pi,k＝{pi,1,k,pi,2,k…pi,N,kEach particle has a state variable of
Figure BDA0002725849740000021
And the existence of variable Ei,j；
Step 3.3, let r be 1, j be 1, calculate the multiple radar weight of each particle in the tracking particle swarm, the concrete steps are:
step 3.3.1, compare E of the target jth particlei,jIf so, entering step 3.3.2, otherwise, calculating the weight of the particle based on the r radar
Figure BDA0002725849740000022
1 and go to step 3.3.4.
Step 3.3.2, calculating the distance of the jth particle under the corresponding r radar
Figure BDA0002725849740000023
Doppler device
Figure BDA0002725849740000024
And orientation value
Figure BDA0002725849740000025
Figure BDA0002725849740000026
Figure BDA0002725849740000027
Figure BDA0002725849740000028
(xr,yr) Representing the r-th radar position.
Step 3.3.3, calculating the weight of the jth particle of the target ith at the moment k based on the ith radar
Figure BDA0002725849740000029
Figure BDA00027258497400000210
Figure BDA00027258497400000211
(m, n, p, k) is at time k, particle
Figure BDA00027258497400000212
The unit position in the radar measurement space; sigmanRepresents the standard deviation, LrDenotes the distance-dependent attenuation constant, LdRepresents a Doppler-dependent attenuation constant, LbShowing the attenuation constant related to the direction, R (m) showing the distance of the target corresponding to the r radar measuring unit, D (n) showing the Doppler of the target corresponding to the r radar measuring unit, B (p) showing the target corresponding to the r radar measuring unitThe orientation value, A, corresponding to the measurement unitr,kThe complex amplitude of the unit is corresponding to the r-th sensor.
Step 3.3.4, if j < N, j equals j +1 and step 3.3.1 is entered, otherwise step 3.3.5 is entered.
Step 3.3.5, if R < R, j is 1, R +1 and step 3.3.1 is entered, otherwise step 3.4 is entered.
Step 3.4, calculating the weight of the multi-radar fusion particle
Figure BDA0002725849740000031
And 3.4.1, making r equal to 1.
Step 3.4.2, normalizing the weight value of the particle corresponding to the r-th radar:
Figure BDA0002725849740000032
step 3.4.3, if R < R, then R ═ R +1 and go to step 3.4.2, otherwise go to step 3.4.4.
Step 3.4.4, calculating the weight of the fused radar particles:
Figure BDA0002725849740000033
Figure BDA0002725849740000034
a multi-radar fusion weight for the jth particle of target i at time k.
Step 3.4.5, normalizing the particle swarm weight after fusion:
Figure BDA0002725849740000035
step 3.5, order
Figure BDA0002725849740000036
Using systematic resampling to track particle swarm Pi,k＝{pi,1,k,pi,2,k…pi,N,kUpdating is carried out;
step 3.6, calculate goal fi,kJudging whether pb is smaller than a found target threshold Myu, if yes, considering the target as a false target, and entering step 3.10, otherwise, considering the target as existing, and entering step 3.7, wherein the target existence probability is:
pb M/N type (10)
Wherein M is the variable E present in the populationi,jA particle number of 1;
step 3.7, re-tracking the target, which comprises the following steps:
step 3.7.1, order
Figure BDA0002725849740000041
According to particle weight
Figure BDA0002725849740000042
Size pair particle Pi,kSorted in descending order.
Step 3.7.2, calculating the state information mean value of the first N/50 particles of the particle swarm
Figure BDA0002725849740000043
Step 3.7.3, randomly creating new particles by taking the state information mean value as a central point, and tracking the particle swarm Pi,kThe state information of the last 0.98 × N particles is replaced, and a particle swarm existence variable Ei,j(j 1, …, N) is set to 1, and the new particle status information is as follows:
Figure BDA0002725849740000044
Figure BDA0002725849740000045
Figure BDA0002725849740000046
Figure BDA0002725849740000047
{xi,j,yi,j}、
Figure BDA0002725849740000048
are respectively Pi,kPosition information and moving speed of the jth particle, Rx,yIs the position radius, RvIs the velocity radius.
And 3.7.4, calculating and tracking the multi-radar weight of each particle in the particle swarm, namely step 3.3.
Step 3.7.5, calculating and calculating particle multi-radar fusion weight
Figure BDA0002725849740000049
See step 3.4.
3.7.6, adopting system resampling to trace particle group pi,k＝{pi,1,k,pi,2,k,…,pi,N,kAre updated, leni＝leni+1, obtain target state estimate
Figure BDA00027258497400000410
I.e. the next moment state of the target.
Step 3.8, if the target i tracks the frame number length leniThe target estimation state is calculated as 3
Figure BDA00027258497400000411
The first track point of the target
Figure BDA00027258497400000412
Difference of (2)
Figure BDA00027258497400000413
Step 3.9, if
Figure BDA00027258497400000414
Let pb be 0, consider the target as a false target, otherwise consider the target as present.
Step 3.10, if i < Tm, i ═ i +1 and go to step 3.2, otherwise go to step 3.9;
3.11, deleting false targets from the target tracking set Taxe, deleting corresponding tracking particle groups, wherein the tracking target set is each target in the Taxe
Figure BDA00027258497400000415
Is updated to
Figure BDA00027258497400000416
Finally obtaining a tracking target set Taxe at the moment k, wherein the number of tracking targets is Tm;
step 4, detecting a new target at the moment k to generate a detection particle swarm
Figure BDA0002725849740000051
For detecting new objects, new objects detected
Figure BDA0002725849740000052
Inputting the detection target set Daxe to obtain a detection target set
Figure BDA0002725849740000053
And detecting a target particle group
Figure BDA0002725849740000054
Wherein h is the h-th target in the detection target set, and specifically is:
step 4.1, making Dm equal to 0; dm is the number of targets in the detection target set;
step 4.2, detecting particle swarm
Figure BDA0002725849740000055
In which each particle is subjected to a state transition to obtain a state variable of each particle as
Figure BDA0002725849740000056
And the existence of variable Eh,jWherein x ish,j,yh,jThe position of the particles in the x, y direction,
Figure BDA0002725849740000057
the speed of the particles in the x and y directions;
step 4.3, calculating the weight of each radar particle in the detection particle swarm
Step 4.3.1, let j equal to 1, r equal to 1, i equal to 1;
4.3.2, calculating the distance between the jth particle and a target i in the combined set of the detection target set Daxe and the tracking target set Taxe and see a formula (11); if it is
Figure BDA0002725849740000058
From the target threshold Jyu, the jth particle weight
Figure BDA0002725849740000059
Setting to 1, and entering a step 4.3.6, otherwise entering a step 4.3.3.
Figure BDA00027258497400000510
xh,j,yh,jFor detecting the position of the particles in the x, y directions, xi,k,yi,kThe position of a target i in the x and y directions is collected for the detection target set and the tracking target set;
step 4.3.3, if i < Tm + Dm, i ═ i +1 and go to step 4.3.2, otherwise go to step 4.3.4.
Step 4.3.4, calculating the distance, Doppler and bearing value of the jth particle based on the r radar:
Figure BDA00027258497400000511
Figure BDA00027258497400000512
Figure BDA00027258497400000513
xr,yrrespectively representing the position of the r-th radar in x, y,
Figure BDA00027258497400000514
respectively representing the distance, Doppler and azimuth values of the jth particle based on the r radar measurement.
Step 4.3.5, calculating the weight of the jth particle based on the r radar measurement
Figure BDA00027258497400000515
Figure BDA0002725849740000061
Figure BDA0002725849740000062
Wherein
Figure BDA0002725849740000063
Representing the weight of the jth particle for the r-th radar, m, n, p, k being particles
Figure BDA0002725849740000064
The position of the cell, σ, in the radar measurement spacenRepresenting standard deviation, R (m) represents the distance of the target corresponding to the r-th radar measurement unit, D (n) represents the Doppler of the target corresponding to the r-th radar measurement unit, B (p) represents the azimuth of the target corresponding to the r-th radar measurement unit, Ar,kRepresenting the complex amplitude of the corresponding cell of the r-th sensor.
Step 4.3.6, if j < N, j equals j +1 and step 4.3.2 is entered, otherwise step 4.3.7 is entered.
In step 4.3.7, if R < R, the process proceeds to step 4.3.2, where j is 1 and R is R + 1.
Step 4.3.8, normalizing the particle swarm weight under the r radar, which is shown in formula (7):
step 4.3.9, calculating the weight of the fused jth particle at the moment k:
Figure BDA0002725849740000065
step 4.4, resampling the particle swarm by adopting a system resampling method;
step 4.5, calculating the detection probability pb of the detected particles according to the formula (10), judging whether pb is smaller than the found target threshold Myu, if yes, going to step 5, otherwise, considering that a new target is detected, and calculating the state estimation of the target
Figure BDA0002725849740000066
Go to step 4.6.
Step 4.6, judging whether the detected target set Daxe and the tracked target set Taxe at the moment k are empty, if so, going to step 5, otherwise, continuously judging whether the new target is a target found in the tracked target set Taxe at the moment k or a target detected in the detected target set Daxe, specifically:
step 4.6.1, changing i to 1;
step 4.6.2, calculating the distance equation (12) of the new target and the target i in the union set of the detection target set and the tracking target set, and judging disi,kIf the value is less than the verification target threshold value Mk, if so, the target is not a new target, the step 4.2 is skipped, the detection particle group is regenerated to detect the new target, otherwise, the step 4.6.3 is entered;
Figure BDA0002725849740000067
where { xi,k，yi,kX is the position of the target i in the x and y directions in the union set of the detection target set and the tracking target set, xD，yDRepresents the x, y direction positions of the new target, respectively;
step 4.6.3, if i < Tm + Dm, i is equal to i +1 and step 4.6.2 is performed, otherwise, step 4.6.4 is performed;
step 4.6.4 New target acquisition detection particle population
Figure BDA0002725849740000071
New object
Figure BDA0002725849740000072
Input to the detected target set
Figure BDA0002725849740000073
In, and setting a new target niTurning to step 4.2, and repeating the steps until no new target is detected, and outputting a detection target set;
and 5, adding the new target into the k-time tracking target set Taxe to obtain an updated k-time tracking target set Taxe ═ f1,k,f2,k,…fTm+DmDetecting particle swarm
Figure BDA0002725849740000074
Updated to track the particle swarm Pi,k。
The invention provides a multi-target particle filter pre-detection tracking method based on target retracing, which is characterized in that in a tracking link, for targets which are tracked and pb of which is greater than a given threshold, tracking particle groups are sorted in a reverse order according to the weight value of particles, and the state mean value of the first N/50 particles is calculated. And generating new particles by taking the state mean value as a central point, replacing state information of the last 0.98 × N particles of the tracking particle swarm, setting the existing values of the particle swarm to be 1, reserving high-quality particles, updating the particle swarm, improving the particle utilization rate, enabling the tracking point trace to be close to a real target quickly, and improving the tracking precision of the target. In addition, the false trace point is enabled to be more quickly close to the real target, and the length len of the trace track of the target is calculatediAnd if the distance between the target and the first track point is 3, the initial state information of the target is inaccurate, the false track is deleted, and the false alarm rate and the tracking error of the target are reduced.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The invention is further analyzed with reference to the following figures.
The invention mainly adopts a computer simulation method for verification, and all the steps are verified correctly on matlab-2016 a. FIG. 1 is a flow chart of the present invention. The specific implementation steps are as follows:
(1) initializing system parameters: the radar scanning period T is 2, the number N of the initialization particles is 3000, the target threshold Myu is 0.7, the particle distance target threshold Jyu is 35, the number Xz of the selected particles is 2, the verification target Mk is 50, and Syu is 0.01.
(2) Obtaining kth time measurements for multiple radars
Figure BDA0002725849740000075
Wherein R is the total number of sensors, m, n, p respectively represent a distance unit, a Doppler unit and an orientation unit, Dr,Dd,DbRespectively range, doppler, and azimuth space cell range.
(3) Taxe ═ f in the tracking target set at the time of k-11,k-1，,f2,k-1…fTm,k-1Target f in (1) }i,k-1Tracking, wherein each target has a tracking particle swarm Pi,k-1＝{pi,1,k-1,pi,2,k-1…pi,N,k-1To the target fi,k-1The tracking process of (1) is as follows:
(a) calculating the number of columns of a tracking target set Taxe, assigning the number of columns to a variable Tm, wherein Tm represents the number of tracking targets;
(b)i＝1，j＝1，r＝1；
(c) the particle group is subjected to state transition to obtain the state of each particle
Figure BDA0002725849740000081
And presence of variable quantity Ei,jWherein x isi,j,yi,jRepresenting the position of the particles in the x, y direction,
Figure BDA0002725849740000082
representing the velocity of the particles in the x, y directions;
(d) if Ei,j<0, then
Figure BDA0002725849740000083
Is 1 and goes to (f);
(e) calculating a weight of a jth particle based on an r-th sensor
Figure BDA0002725849740000084
Wherein xr,yrRespectively representing the position of the r-th radar in x, y,
Figure BDA0002725849740000085
respectively representing the distance, Doppler and orientation values of the jth particle under the ith particle, R (m) representing the distance of the target corresponding to the ith radar measuring unit, D (n) representing the Doppler of the target corresponding to the ith radar measuring unit, B (p) representing the orientation value of the target corresponding to the ith radar measuring unit, Ar,kRepresenting the complex amplitude of the corresponding cell of the r-th sensor. .
Figure BDA0002725849740000086
Figure BDA0002725849740000087
Figure BDA0002725849740000088
Figure BDA0002725849740000089
Figure BDA00027258497400000810
(f) If j < N, j equals j +1 and goes to (d), otherwise goes to (g);
(g) if R < R, let j equal 1, R equal R +1 and go to (d), otherwise go to (h);
(h) let r be 1;
(i) normalizing the particle swarm weight under the r radar:
Figure BDA0002725849740000091
(j) if R < R, then R ═ R +1 and enter (i), otherwise enter (k);
(k) calculating the weight after the fusion of the particles, wherein
Figure BDA0002725849740000092
Represents the weight power value of the particles under the r radar:
Figure BDA0002725849740000093
(l) Order to
Figure BDA0002725849740000094
Obtaining tracking particle swarm p at k moment by adopting system resampling methodi,k＝{pi,1,k,pi,2,k…pi,N,k}；
(m) calculating the target fi,k-1M/N, M being the presence variable Ei,jIf pb is less than the find target threshold Myu, then the target is considered a false target and enters (u), if more than the threshold Myu the target exists, then enters (n);
(n) order
Figure BDA0002725849740000095
According to particle weight
Figure BDA0002725849740000096
Size pair particle Pi,kAnd (5) sorting in a descending order.
(o)Calculating the state information mean value of the first N/50 particles of the particle swarm
Figure BDA0002725849740000097
(p) with the mean value of the state information as the central point, creating 0.98 × N new particles, replacing and tracking 0.98 × N particles after the particle swarm, and enabling the particle swarm to have a variable Ei,j(j ═ 1, …, N) is all 1, and the replacement particle state information is as follows:
Figure BDA0002725849740000098
Figure BDA0002725849740000099
Figure BDA00027258497400000910
Figure BDA00027258497400000911
(q) calculating the multiple radar weights of each particle in the tracking particle swarm, see steps (c) - (j).
(r) calculating and calculating particle radar fusion weight
Figure BDA00027258497400000912
See step (k).
(s) using systematic resampling to track the population Pi,k＝{pi,1,k,pi,2,k,…,pi,N,kAre updated, leni＝leni+1, obtain target state estimate
Figure BDA00027258497400000913
(t) if target i tracks frame number length leniThe target estimation state f is calculated as 3i,kDifference from target detection stateValue of
Figure BDA0002725849740000101
If it is
Figure BDA0002725849740000102
Setting pb to 0, considering the target as a false target, and otherwise, setting pb to 1, considering the target as present;
(u) if i < Tm, then i ═ i +1 and enter (c), otherwise enter (v);
(v) deleting the target from the target tracking set Taxe and deleting the tracking particle swarm Pi,kTracking the target set as each target f in Taxei,k-1Is updated to fi,kFinally, a tracking target set Taxe at the moment k is obtained;
(w) updating the number of columns in the target tracking set Taxe and updating the number of targets Tm in the target tracking set;
(4) detecting a new target by using the detection particle group, and starting the new target number Dm to be 0, wherein the process is as follows:
(a) carrying out state transition on the detection particle swarm to obtain the state of each particle as
Figure BDA0002725849740000103
And the existence of variable Eh,jWherein x ish,j,yh,jThe position of the particles in the x, y direction,
Figure BDA0002725849740000104
is the x, y direction velocity of the particle;
(b) let r be 1, j be 1, i be 1, and target flag alloxit be 0;
(c) calculating the distance between the jth particle and a target i in a union set of a detection target set Daxe and a tracking target set Taxe:
Figure BDA0002725849740000105
(d) if it is
Figure BDA0002725849740000106
Then alloxit is 0 and the weight of the jth particle under the r-th radar
Figure BDA0002725849740000107
Setting 1 and switching to (g), otherwise, switching to (e) when alloxit is 1;
(e) if i < Tm + Dm, i ═ i +1 and enter (c), otherwise enter (f);
(f) calculating the weight of the jth particle corresponding to the r radar
Figure BDA0002725849740000108
Figure BDA0002725849740000109
Figure BDA00027258497400001010
Figure BDA00027258497400001011
Figure BDA00027258497400001012
Figure BDA0002725849740000111
(g) If j < N, j equals j +1, i equals 1, and go to (c), otherwise go to (h);
(h) if R < R, let j equal 1, i equal 1, R equal R +1 and go to (c), otherwise go to (i);
(i)r＝1；
(j) normalizing the particle swarm weight under the r radar:
Figure BDA0002725849740000112
(k) if R < R, then R +1 and go to (j), otherwise go to (l);
(l) Fusing the particle swarm weights:
Figure BDA0002725849740000113
(M) calculating the probability pb of detection of the detected particles, pb being M/N, M being the presence variable Eh,jA particle number of 1;
(n) if pb is less than the threshold go to (5), otherwise a new target is detected, calculating a state estimate for the target
Figure BDA0002725849740000114
Go to (o).
(o) if Tm + Td is equal to 0, go to (5), otherwise go to (p) to determine whether the newly detected target is the found target;
(p) let i be 1 and new target flag be 0;
(q) calculating the distance between the newly detected target and the target i in the detection target set and the tracking target set as follows:
Figure BDA0002725849740000115
where { xq，yqTarget fi,kX, y direction of (c) { xD，yDThe position of the newly detected target in the x and y directions is used as the position of the target;
(r) if disi,k<If the Mk is not the new target, if the flag is 1, jumping to the step (a), regenerating a detection particle group to detect the new target, and otherwise, entering the step(s);
(s) if i < Tm + Dm, i ═ i +1, proceed to step (q), otherwise proceed to (t);
(t) if flag is 0, obtaining a detection particle group as a new target
Figure BDA0002725849740000121
New object
Figure BDA0002725849740000122
Input to the detected target set
Figure BDA0002725849740000123
When the Dm is equal to Dm +1, turning to the step (a), regenerating a detection particle group to detect a new target, and circulating until the new target cannot be detected, and outputting a detection target set;
(5) adding the detection target set Daxe in the period into the tracking target set Taxe to obtain an updated tracking target set Taxe { f }1,k,f2,k,…fTm+DmDetecting particle swarm
Figure BDA0002725849740000124
Updated to track the particle swarm Pi,k；
Simulation scene:
the total 5 radars are all positioned at the original point, under the condition that the number of the targets is 5, the target signal-to-noise ratios are respectively 6dB, 9dB and 12dB, the targets are far away from the radars, and under the condition that the target signal-to-noise ratios are large in difference, the target re-tracking-based multi-target particle filter pre-detection tracking method reduces the target false alarm rate and improves the target early-stage tracking precision.

Claims (1)

1. A multi-target particle filter pre-detection tracking method based on target re-tracking is characterized by comprising the following steps:
step 1, initializing parameters: radar scanning period T, total observation frame number K, number of particles in particle swarm N, radar number R, frame length len of target tracking trackiDistance, Doppler and azimuth space cell distance are Dr,Dd,DbError threshold Error;
step 2, reading the k frame measurement of multiple radars
Figure FDA0002725849730000011
Wherein the content of the first and second substances,
Figure FDA0002725849730000012
the measurement in a measurement unit (m, n, p) of the echo data of the kth frame of the r-th radar is shown, wherein m, n and p respectively show a distance unit, a Doppler unit and a direction unit;
step 3, setting the tracking target set Taxe at the moment k-1 as f1,k-1，,f2,k-1…fTm,k-1Tracking the targets in the data, wherein Tm is the number of the targets in the tracking set, and each target fi,k-1All have a tracking particle swarm Pi,k-1＝{pi,1,k-1,pi,2,k-1…pi,N,k-1In which p isi,j,k-1The jth particle representing the ith target time k-1;
step 3.1, changing i to 1;
step 3.2, tracking particle swarm P of ith targeti,k-1＝{pi,1,k-1,pi,2,k-1…pi,N,k-1Performing state transition on each particle in the target particle swarm P to obtain a target tracking particle swarm Pi,k＝{pi,1,k,pi,2,k…pi,N,kEach particle has a state variable of
Figure FDA0002725849730000013
And the existence of variable Ei,j；
Step 3.3, let r be 1, j be 1, calculate the multiple radar weight of each particle in the tracking particle swarm, the concrete steps are:
step 3.3.1, compare E of the target jth particlei,jIf so, entering step 3.3.2, otherwise, calculating the weight of the particle based on the r radar
Figure FDA0002725849730000014
1 and go to step 3.3.4;
step 3.3.2, calculating the distance of the jth particle under the corresponding r radar
Figure FDA0002725849730000015
Doppler device
Figure FDA0002725849730000016
And orientation value
Figure FDA0002725849730000017
Figure FDA0002725849730000018
Figure FDA0002725849730000019
Figure FDA00027258497300000110
(xr,yr) Represents the r-th radar position;
step 3.3.3, calculating the weight of the jth particle of the target ith at the moment k based on the ith radar
Figure FDA0002725849730000021
Figure FDA0002725849730000022
Figure FDA0002725849730000023
(m, n, p, k) is at time k, particle
Figure FDA0002725849730000024
The unit position in the radar measurement space; sigmanRepresents the standard deviation, LrDenotes the distance-dependent attenuation constant, LdRepresents a Doppler-dependent attenuation constant, LbIndicating orientation-dependent attenuationNumber, R (m) represents the distance of the target corresponding to the r-th radar measurement unit, D (n) represents the Doppler of the target corresponding to the r-th radar measurement unit, B (p) represents the azimuth of the target corresponding to the r-th radar measurement unit, Ar,kThe complex amplitude of the unit corresponding to the r-th sensor is obtained;
step 3.3.4, if j < N, j ═ j +1 and go to step 3.3.1, otherwise go to step 3.3.5;
step 3.3.5, if R < R, j is 1, R +1 and go to step 3.3.1, otherwise go to step 3.4;
step 3.4, calculating the weight of the multi-radar fusion particle
Figure FDA0002725849730000025
Step 3.4.1, making r equal to 1;
step 3.4.2, normalizing the weight value of the particle corresponding to the r-th radar:
Figure FDA0002725849730000026
step 3.4.3, if R < R, then R ═ R +1 and go to step 3.4.2, otherwise go to step 3.4.4;
step 3.4.4, calculating the weight of the fused radar particles:
Figure FDA0002725849730000027
Figure FDA0002725849730000028
a multi-radar fusion weight for a jth particle of a target i at time k;
step 3.4.5, normalizing the particle swarm weight after fusion:
Figure FDA0002725849730000031
step 3.5, order
Figure FDA0002725849730000032
Using systematic resampling to track the particle swarm pi,k＝{pi,1,k,pi,2,k,…,pi,N,kUpdating is carried out;
step 3.6, calculate goal fi,kJudging whether pb is smaller than a found target threshold Myu, if yes, considering the target as a false target, and entering step 3.10, otherwise, considering the target as existing, and entering step 3.7, wherein the target existence probability is:
pb M/N type (10)
Wherein M is the variable E present in the populationi,jA particle number of 1;
step 3.7, re-tracking the target, which comprises the following steps:
step 3.7.1, order
Figure FDA0002725849730000033
According to particle weight
Figure FDA0002725849730000034
Size pair particle Pi,kSorting according to descending order;
step 3.7.2, calculating the state information mean value of the first N/50 particles of the particle swarm
Figure FDA0002725849730000035
Step 3.7.3, randomly creating new particles by taking the state information mean value as a central point, and tracking the particle swarm Pi,kThe state information of the last 0.98 × N particles is replaced, and a particle swarm existence variable Ei,jSetting 1, j to 1, …, N, and establishing the new particle state information as follows:
Figure FDA0002725849730000036
Figure FDA0002725849730000037
Figure FDA0002725849730000038
Figure FDA0002725849730000039
{xi,j,yi,j}、
Figure FDA00027258497300000310
are respectively Pi,kPosition information and moving speed of the jth particle, Rx,yIs the position radius, RvIs the velocity radius;
step 3.7.4, calculating and tracking the multi-radar weight of each particle in the particle swarm, see step 3.3;
step 3.7.5, calculating and calculating particle multi-radar fusion weight
Figure FDA00027258497300000311
See step 3.4;
3.7.6, adopting system resampling to trace particle group pi,k＝{pi,1,k,pi,2,k,…,pi,N,kAre updated, leni＝leni+1, obtain target state estimate
Figure FDA0002725849730000041
The state of the target at the next moment is obtained;
step 3.8, if the target i tracks the frame number length leniThe target estimation state is calculated as 3
Figure FDA0002725849730000042
The first track point of the target
Figure FDA0002725849730000043
Difference of (2)
Figure FDA0002725849730000044
Step 3.9, if
Figure FDA0002725849730000045
Let pb be 0, consider the target as a false target, otherwise consider the target as present;
step 3.10, if i < Tm, i ═ i +1 and go to step 3.2, otherwise go to step 3.11;
3.11, deleting false targets from the target tracking set Taxe and deleting corresponding tracking particle swarm, wherein the tracking target set estimates the state of each target in the Taxe
Figure FDA0002725849730000046
Is updated to
Figure FDA0002725849730000047
Finally obtaining a tracking target set Taxe at the moment k, wherein the number of tracking targets is Tm;
step 4, detecting a new target at the moment k to generate a detection particle swarm
Figure FDA0002725849730000048
For detecting new objects, new objects detected
Figure FDA0002725849730000049
Inputting the detection target set Daxe to obtain a detection target set
Figure FDA00027258497300000410
And detecting a target particle group
Figure FDA00027258497300000411
Wherein h is the h-th target in the detection target set, and specifically is:
step 4.1, making Dm equal to 0; dm is the number of targets in the detection target set;
step 4.2, detecting particle swarm
Figure FDA00027258497300000412
In which each particle is subjected to a state transition to obtain a state variable of each particle as
Figure FDA00027258497300000413
And the existence of variable Eh,jWherein x ish,j,yh,jThe position of the particles in the x, y direction,
Figure FDA00027258497300000414
the speed of the particles in the x and y directions;
step 4.3, calculating the weight of each radar particle in the detection particle swarm
Step 4.3.1, let j equal to 1, r equal to 1, i equal to 1;
4.3.2, calculating the distance between the jth particle and a target i in the combined set of the detection target set Daxe and the tracking target set Taxe and see a formula (11); if it is
Figure FDA00027258497300000415
Then the jth particle weight
Figure FDA00027258497300000416
Setting as 1, entering a step 4.3.6, otherwise entering a step 4.3.3;
Figure FDA00027258497300000417
xh,j,yh,jfor detecting the position of the particles in the x, y directions, xi,k,yi,kThe position of a target i in the x and y directions is collected for the detection target set and the tracking target set;
step 4.3.3, if i < Tm + Dm, i ═ i +1 and go to step 4.3.2, otherwise go to step 4.3.4;
step 4.3.4, calculating the distance, Doppler and bearing value of the jth particle based on the r radar:
Figure FDA0002725849730000051
Figure FDA0002725849730000052
Figure FDA0002725849730000053
xr,yrrespectively representing the position of the r-th radar in x, y,
Figure FDA0002725849730000054
respectively representing the distance, Doppler and azimuth values of the jth particle based on the r radar measurement;
step 4.3.5, calculating the weight of the jth particle based on the r radar measurement
Figure FDA0002725849730000055
Figure FDA0002725849730000056
Figure FDA0002725849730000057
Wherein
Figure FDA0002725849730000058
Representing the weight of the jth particle for the r-th radar, m, n, p, k being particles
Figure FDA0002725849730000059
The position of the cell, σ, in the radar measurement spacenRepresenting standard deviation, R (m) represents the distance of the target corresponding to the r-th radar measurement unit, D (n) represents the Doppler of the target corresponding to the r-th radar measurement unit, B (p) represents the azimuth of the target corresponding to the r-th radar measurement unit, Ar,kRepresenting the complex amplitude of the corresponding unit of the r sensor;
step 4.3.6, if j < N, j ═ j +1 and go to step 4.3.2, otherwise go to step 4.3.7;
step 4.3.7, if R < R, let j equal to 1, R equal to R +1 and go to step 4.3.2;
step 4.3.8, normalizing the particle swarm weight under the r radar, which is shown in formula (7):
step 4.3.9, calculating the weight of the fused jth particle at the moment k:
Figure FDA00027258497300000510
step 4.4, resampling the particle swarm by adopting a system resampling method;
step 4.5, calculating the detection probability pb of the detected particles according to the formula (10), judging whether pb is smaller than the found target threshold Myu, if yes, going to step 5, otherwise, considering that a new target is detected, and calculating the state estimation of the target
Figure FDA0002725849730000061
Entering the step 4.6;
step 4.6, judging whether the detected target set Daxe and the tracked target set Taxe at the moment k are empty, if so, going to step 5, otherwise, continuously judging whether the new target is a target found in the tracked target set Taxe at the moment k or a target detected in the detected target set Daxe, specifically:
step 4.6.1, changing i to 1;
step 4.6.2, calculating the sum of the new target and the detected target set and the tracked target setSee equation (12) for the distance of the concentrated target i, and determine disi,kIf the value is less than the verification target threshold value Mk, if so, the target is not a new target, the step 4.2 is skipped, the detection particle group is regenerated to detect the new target, otherwise, the step 4.6.3 is entered;
Figure FDA0002725849730000062
where { xi,k，yi,kX is the position of the target i in the x and y directions in the union set of the detection target set and the tracking target set, xD，yDRepresents the x, y direction positions of the new target, respectively;
step 4.6.3, if i < Tm + Dm, i is equal to i +1 and step 4.6.2 is performed, otherwise, step 4.6.4 is performed;
step 4.6.4 New target acquisition detection particle population
Figure FDA0002725849730000063
New object
Figure FDA0002725849730000064
Input to the detected target set
Figure FDA0002725849730000065
In, and setting a new target niTurning to step 4.2, and repeating the steps until no new target is detected, and outputting a detection target set;
and 5, adding the new target into the k-time tracking target set Taxe to obtain an updated k-time tracking target set Taxe ═ f1,k,f2,k,...fTm+DmDetecting particle swarm
Figure FDA0002725849730000066
Updated to track the particle swarm Pi,k。
CN202011102472.XA 2020-10-15 2020-10-15 Multi-target particle filtering pre-detection tracking method based on target re-tracking Active CN112415503B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011102472.XA CN112415503B (en) 2020-10-15 2020-10-15 Multi-target particle filtering pre-detection tracking method based on target re-tracking

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011102472.XA CN112415503B (en) 2020-10-15 2020-10-15 Multi-target particle filtering pre-detection tracking method based on target re-tracking

Publications (2)

Publication Number Publication Date
CN112415503A true CN112415503A (en) 2021-02-26
CN112415503B CN112415503B (en) 2023-10-20

Family

ID=74854607

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011102472.XA Active CN112415503B (en) 2020-10-15 2020-10-15 Multi-target particle filtering pre-detection tracking method based on target re-tracking

Country Status (1)

Country Link
CN (1) CN112415503B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116167594A (en) * 2023-04-21 2023-05-26 中国地质大学（北京） Unmanned aerial vehicle platform for detecting vital signs of human body under landslide body and detection method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014190959A (en) * 2013-03-28 2014-10-06 Mitsubishi Electric Corp Target tracking device
CN107783104A (en) * 2017-10-17 2018-03-09 杭州电子科技大学 Tracking before a kind of more asynchronous sensor single goals detection based on particle filter
CN110376556A (en) * 2019-06-11 2019-10-25 杭州电子科技大学 A kind of preceding tracking of two-layered spherical particle filtering detection based on algorithm of tournament selection
CN111007454A (en) * 2019-10-28 2020-04-14 重庆邮电大学 Extended target tracking method based on cooperative target information

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014190959A (en) * 2013-03-28 2014-10-06 Mitsubishi Electric Corp Target tracking device
CN107783104A (en) * 2017-10-17 2018-03-09 杭州电子科技大学 Tracking before a kind of more asynchronous sensor single goals detection based on particle filter
CN110376556A (en) * 2019-06-11 2019-10-25 杭州电子科技大学 A kind of preceding tracking of two-layered spherical particle filtering detection based on algorithm of tournament selection
CN111007454A (en) * 2019-10-28 2020-04-14 重庆邮电大学 Extended target tracking method based on cooperative target information

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
张婷;许可;: "一种利用点目标响应和加权窗进行海洋雷达高度计重跟踪的新方法", 电子学报, no. 03 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116167594A (en) * 2023-04-21 2023-05-26 中国地质大学（北京） Unmanned aerial vehicle platform for detecting vital signs of human body under landslide body and detection method
CN116167594B (en) * 2023-04-21 2023-10-17 中国地质大学（北京） Unmanned aerial vehicle platform for detecting vital signs of human body under landslide body and detection method

Also Published As

Publication number Publication date
CN112415503B (en) 2023-10-20

Similar Documents

Publication Publication Date Title
CN111812636B (en) Particle filtering pre-detection tracking method based on weight fusion selection
CN1940591A (en) System and method of target tracking using sensor fusion
CN110865343B (en) LMB-based particle filter tracking-before-detection method and system
KR101628154B1 (en) Multiple target tracking method using received signal strengths
CN104715154B (en) Core K average Data Associations based on KMDL criterion criterions
JP6773606B2 (en) Radar device and its radar signal processing method
CN107656265A (en) Particle filter fusion method for tracking short flight path before multi frame detection
CN109858526A (en) Sensor-based multi-target track fusion method in a kind of target following
CN104181524A (en) Particle-number-adaptive multi-target particle filtering tracking-before-detecting method
CN111830501B (en) HRRP history feature assisted signal fuzzy data association method and system
CN112379393B (en) Train collision early warning method and device
CN111562570A (en) Vehicle sensing method for automatic driving based on millimeter wave radar
CN112313536B (en) Object state acquisition method, movable platform and storage medium
CN110376556B (en) Double-layer particle filter track-before-detect method based on tournament selection
CN112415503A (en) Multi-target particle filter pre-detection tracking method based on target re-tracking
CN114137526A (en) Label-based vehicle-mounted millimeter wave radar multi-target detection method and system
CN110411499B (en) Evaluation method and evaluation system for detection and identification capability of sensor
CN111257826B (en) Multi-source heterogeneous sensor composite tracking method
CN110824478B (en) Automatic classification method and device for precipitation cloud types based on diversified 3D radar echo characteristics
CN115327529A (en) 3D target detection and tracking method fusing millimeter wave radar and laser radar
CN114119465A (en) Point cloud data processing method and device
CN112380986A (en) Vehicle type detection method and system for resisting complex street background interference and storage medium
CN115220002B (en) Multi-target data association tracking method and related device for fixed single station
CN113052871A (en) Target detection and automatic tracking algorithm based on intelligent selection strategy
De Candido et al. An analysis of distributional shifts in automated driving functions in highway scenarios

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