CN106443622B - A kind of distributed object tracking based on improvement joint probability data association - Google Patents
A kind of distributed object tracking based on improvement joint probability data association Download PDFInfo
- Publication number
- CN106443622B CN106443622B CN201610821318.5A CN201610821318A CN106443622B CN 106443622 B CN106443622 B CN 106443622B CN 201610821318 A CN201610821318 A CN 201610821318A CN 106443622 B CN106443622 B CN 106443622B
- Authority
- CN
- China
- Prior art keywords
- target
- echo
- probability
- sensor
- state estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- 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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to Distributed filtering and target following technical fields, and in particular to a kind of based on the distributed object tracking for improving joint probability data association.The present invention includes: that (1) establishes motion model to each target respectively to sensor multiple-target system;(2) data correlation filtering method is carried out to each sensor respectively, obtains state estimation of the respective sensor about the target in detection range;(3) respectively the state estimation of sensor carries out space correlation, and the state estimation for then merging same target obtains final Target state estimator value.Simulation results show the performance of the algorithm, can reach the precision as classical joint improvement version data correlation method, and be better than joint probability data association method on time performance, reduce the time of algorithm.
Description
Technical field
The invention belongs to Distributed filtering and target following technical fields, and in particular to one kind is based on improvement joint probability number
According to associated distributed object tracking.
Background technique
The ambient enviroment problem that the research of simple target tracking considers is fairly simple, measures and believes from single goal single-sensor
The research of single goal multi-sensor information is arrived in the research of breath, and comprehensive more, more fully measurement information carries out target following, to
Obtain high-precision target status information.Between the complexity and variability of modern external environment, gradually to target following research
Single target tracking in the monocycle that conforms to the principle of simplicity border develops to the tracing mode under complex environment and target-rich environment.Sensor at this time
The tracked target information of measurement may be influenced by from clutter, other targets, and the source of measurement information, which not can determine that, to be come
From which specific target, the Target Tracking Problem being directed under these complex situations needs to solve sensing data source first
The problem of, the measurement information which is only real target specified in sensor information.Data association algorithm be exactly be solve it is this kind of
What problem occurred, establish the relationship between sensor measurement information and target.So tracking ring at multiple target (or dense clutter)
Under border, the committed step of object tracking process first is that determining the source of observation data, i.e. data correlation.Current existing data
Correlating method has closest to method (Nearest Neighbor, NN), probability data correlation method (Probability Data
Association, PDA), joint probability data association method (Joint Probability Data Association,
JPDA), more subjunctives (Multiple Hypothesis Tracking, MHT) and broad sense multidimensional distribution method etc..With new
The development of technology, some scholars are dedicated to being used for neural network, fuzzy logic technology etc. to solve the problems, such as data correlation.
Joint probability data association (Joint Probability Data Association, JPDA) method is more at present
More satisfactory a kind of algorithm in the data association algorithm of target following, but the quantity of joint event is with the increase of number of echoes
The case where being exponentially increased causes the calculation amount of joint probability to greatly increase, and is not able to satisfy requirement of the track algorithm to real-time,
So the calculation amount for reducing algorithm is one of main improvement direction of JPDA.
Summary of the invention
The purpose of the present invention is to solve the above problems, provide a kind of based on point for improving joint probability data association
Cloth method for tracking target.
The object of the present invention is achieved like this:
(1) motion model is established to each target respectively to sensor multiple-target system;
(2) data correlation filtering method is carried out to each sensor respectively, obtains respective sensor about in detection range
Target state estimation
Data correlation filtering method includes:
(2.1) tracking gate of each target is set, determines effective echo m of targetk, and calculate initial association probability, structure
Make determining matrixJ=1,2 ... mkExpression echo, t=1,2 ... T indicates target;Determine the element in matrix
It only indicates the relationship between target and effective echo, does not consider the relationship of echo and false-alarm;Each target is first calculated according to PDA method
Echo association probability in tracking gate
Effective echo j is derived from the probability of target t:
There is no echo to be derived from the probability of target t:
bt=λ (1-PDPG)|2πSt(k)|1/2,Wherein parameter PDExpression pair
The detection probability of real goal, PGIndicate that the measurement of sensor falls into the probability in tracking gate, λ indicates the density of environment clutter, St
(k)=H (k) Pt(k|k-1)HT(k)+R (k) is the new breath covariance of target,Indicate a certain
Measure the residual error to target;
(2.2) according to the public echo ensembles Pub for determining that matrix Ω determines moment k;
(2.3) to association probabilityIt is modified;
By the formula of (2.2), the echo in public echo ensembles Pub is associated with two or more targets, according to echo
Practical utilization power guarantees that the utilization rate of public echo j is no more than 1, association probability is handled as follows:
For not common echo, association probability does not influence, and does not have to amendment,
When in tracking gate without echo, there is corresponding association probabilityAssociation probability βjt(k) pact of following formula need to be met
Beam condition, makes the association probability of target and for 1,
To target association probabilityCarry out following normalized:
The association probability of echo is calculatedCorresponding state estimation covariance is obtained,
(3) respectively the state estimation of sensor carries out space correlation, and the state estimation for then merging same target obtains most
Whole Target state estimator value.
The step (1) specifically:
Motion model is established to each target respectively to sensor multiple-target system;Establish Nonlinear Parameter tracking system
System model:
For target following since the quantity of state and measurement of selection are under different coordinate systems, target following is one
A nonlinear state estimation problem;W in formulak、VkThe measurement noise of system noise and respective sensor is respectively indicated, and full
Sufficient condition:
WhereinIndicate k moment sensor i to the measurement information of target;During target following, examined at sensor
The information measured calculates the association probability that all echoes are derived from target by data association algorithm, then calculates corresponding state and estimates
Meter.
The step (3) are as follows:
The state estimation of respective sensor carries out space correlation, then merge same target state estimation obtain it is final
Target state estimator value;
Wherein, ci,lIt indicates the diffusion weight of each sensor, and meets
The beneficial effects of the present invention are: simulation results show the performance of the algorithm, can reach and classical joint is improved
The same precision of version data correlation method, and it is better than joint probability data association method on time performance, reduce algorithm
Time.
Detailed description of the invention
Fig. 1 is tracking situation of the distribution UKF-IJPDA algorithm to three targets.
Fig. 2 is position RMSE curve of the distribution UKF-IJPDA algorithm to three targets.
Fig. 3 is the position of the distributed UKF algorithm based on IJPDA and the distributed UKF algorithm based on JPDA to target 1
RMSE curve comparison.
Fig. 4 is the position of the distributed UKF algorithm based on IJPDA and the distributed UKF algorithm based on JPDA to target 2
RMSE curve comparison.
Fig. 5 is the position of the distributed UKF algorithm based on IJPDA and the distributed UKF algorithm based on JPDA to target 3
RMSE curve comparison.
Specific embodiment
Below in conjunction with attached drawing, the present invention is described in further detail.
Fig. 1 is tracking situation of the distribution UKF-IJPDA algorithm to three targets, gives the reality of 3 targets movement
Track and estimation geometric locus.Wherein target 1 does crisscross motion with target 2, target 3 respectively, and target 2 does parallel fortune with target 3
It is dynamic.It can be seen that estimation geometric locus is offset slightly from actual path, but simultaneously near the track cross position of two targets
Do not occur the case where with wrong target;Especially for two targets for doing parallel motion, state estimation does not cause confusion
The case where, estimation curve can keep up with actual path always, be not in crossing instances.The experimental results showed that distribution UKF-
The performance of IJPDA algorithm can accurately guarantee the tracking effect of target.
Fig. 2 describes distributed UKF-IJPDA algorithm to the position RMSE curve of three targets.It can be seen that with emulation
The progress of experimentation, the position RMSE of each target finally converge to a stable value, illustrate the algorithm in multiple target
There is preferable tracking performance in tracking.
Fig. 3~Fig. 5 respectively describes the distributed UKF algorithm based on IJPDA and the distributed UKF algorithm based on JPDA closes
In the position RMSE curve comparison of three targets.It is found that the distributed UKF-IJPDA algorithm of this paper can reach and based on JPDA's
The same estimated accuracy of distributed UKF algorithm, the position RMSE of two kinds of algorithms are not much different.
For multi-sensor multi-target tracking, each sensor can detect the measurement information of multiple targets, need at this time
Sensing data is merged.Shape by the way of Distributed Database cluster by respective sensor about the same target
State estimation fusion, the available end-state about target are estimated.Distribution UKF-IJPDA method proposed in this paper be exactly
It is realized on the basis of state layer fusion, IJPDA filter procedure is carried out to each sensor first, obtains corresponding mesh
State estimation is marked, then the state estimation for the same target that different sensors obtain is melted according to the strategy that diffusion type merges
It closes, obtains comprehensive Target state estimator.
The specific calculating process of distribution UKF-IJPDA algorithm proposed in this paper is as follows.
Step 1: motion model is established respectively to each target in sensor multiple-target system.
Step 2: improving data correlation filtering method to each sensor respectively, obtains respective sensor about inspection
Survey the state estimation of the target in range
Improved data correlation filtering method (IJPDA algorithm) is as follows:
When calculating the association probability of multiple tracking target conditions with PDA algorithm, such as when the public echo j distance of target
Target t1、t2Apart from it is suitable when, PDA algorithm calculate weightIt is all larger, i.e., the echo to the tracks of two targets more
New capital is weighted again.But actually public echo j is only possible to from a target (t1Or t2), when the power to a target
It is necessarily lower to the power of other targets when heavier.The case where for the competition of this echo, following vacation is made with target to measuring
If:
All targets relevant for public echo are all related to echo, and only probability is different, and meets echo to target
Association probability and be 1.
All echoes in target following door are all related to target, and only probability is different and to meet the institute of target relevant general
Rate and be 1.
At this time in IJPDA algorithm, the calculating process of association probability is as follows.
(1): the tracking gate of each target being set, determines effective echo m of targetk, and calculate initial association probability.Construction
Determine matrixJ=0,1 ..., mkIndicate that echo, t=1,2 ..., T indicate target.It determines in matrix at this time
Element only indicate the relationship between target and effective echo, do not consider the relationship of echo and false-alarm.First calculated according to PDA method
Echo association probability in each target following door
Effective echo j is derived from the probability of target t:
There is no echo to be derived from the probability of target t:
In upper two formula, bt=λ (1-PDPG)|2πSt(k)|1/2,Wherein join
Number PDIndicate the detection probability to real goal, PGIndicate that the measurement of sensor falls into the probability in tracking gate, λ indicates that environment is miscellaneous
The density of wave, St(k)=H (k) Pt(k|k-1)HT(k)+R (k) is the new breath covariance of target,
Indicate a certain residual error measured to target.
(2): according to the public echo ensembles Pub for determining that matrix Ω determines moment k;
(3): to association probabilityAmendment.
Known by formula (3), the echo in public echo ensembles Pub is associated with two or more targets, when echo is apart from these
When the tracking gate center of target is nearest, the association probability obtained by PDA algorithm is all very big, all to the state estimation of target
Biggish weight is contributed, which is reused by multiple targets simultaneously.But in fact, an echo can only be associated with a mesh
Mark, when the more newly assigned weight of the state to some target is more, necessarily reduces the weight of other Target Assignments, otherwise
There is the case where echo recycling, it is therefore desirable to be modified to the public echo association probability that above formula acquires.According to echo
Practical utilization power, guarantee public echo j utilization rate be no more than 1, association probability can be handled as follows:
For not common echo, association probability does not influence, and does not have to amendment,
All echoes in target following door are all related to the target, and contribute the state estimation of the target certain power
Value.When in tracking gate without echo, there is corresponding association probabilityFor the stability for guaranteeing system, association probability βjt(k)
The constraint condition that following formula need to be met makes the association probability of target and for 1.
Therefore to target association probabilityCarry out following normalized:
The association probability of echo is thus calculatedIt can be obtained by corresponding state by formula (7) and formula (8) to estimate
Covariance is counted, to prepare for estimation in next step.
Step 3: the state estimation of respective sensor carries out space correlation, and the state estimation for then merging same target obtains
To final Target state estimator value.
A kind of distributed object tracking based on improvement joint probability data association of the invention, including it is following
Step:
Step 1: motion model is established to each target respectively to sensor multiple-target system.Specifically, establishing non-
Linear goal tracking system model is as follows:
WhereinIndicate k moment sensor i to the measurement information of target.
For target following since the quantity of state and measurement of selection are under different coordinate systems, target following is one
A nonlinear state estimation problem.W in formulak、VkThe measurement noise of system noise and respective sensor is respectively indicated, and full
Sufficient condition:
During target following, the information that detects at sensor is simultaneously not all measurement letter about some target
Breath may include some clutter information, or the measurement information about other targets.It needs to pass through data association algorithm at this time
The association probability that all echoes are derived from target is calculated, corresponding state estimation is then calculated.
Step 2: carrying out data correlation filtering method to each sensor respectively, obtains respective sensor about detection model
The state estimation of target in enclosing
(1): setting target following door, and confirm the validity echo number mk, obtain determining matrix
(2): according to the state estimation and covariance of last moment, by UKF method to each target carry out state update and
It measures and updates.Wm、WcThe weight of sampling policy respectively in UKF algorithm.ByIt calculates sampled point and obtains predicted value
ByCalculate sampled pointAnd obtain nonlinear function propagation values
The state estimation of each target is calculated according to IJPDA method:
Indicate k moment echo j (i.e. measurement Zk(j)) to the state estimation of target t, such as formula:
Step 3: the state estimation of respective sensor carries out space correlation, and the state estimation for then merging same target obtains
To final Target state estimator value.
Wherein, ci,lIt indicates the diffusion weight of each sensor, and meets
Embodiment: being based on radar sensor, emulation experiment is carried out to the distributed UKF-IJPDA algorithm of proposition, with verifying
The performance of algorithm.
Assumed condition is as follows: having 3 targets, 10 sensor nodes, each sensor can obtain the position of 3 targets
Information, each target do the linear motion of nearly constant speed.The position made even in the two-dimensional Cartesian coordinate system corresponding coordinate axis direction in face
The state with speed for target is set, state CV model is established, using the radial distance of radar sensor and azimuth as observation, builds
Vertical sensor model, shown in obtained trace model such as formula (22):
Wherein (bx,by) be sensor position, each sensor works independently, and assumes that all the sensors noise variance is special
Property consistent, R=diag { 100 0.04 }, Q=diag { 0.01 0.01 }.Clutter obedience in tracking gate is uniformly distributed, and clutter
Density is λ=2 × 10-6/m2, target detection probability PD=1, door probability PG=0.99, the threshold value of oval tracking gate be γ=
9.21;Sampling time is 50s, sampling interval 1s, carries out 50 Monte-Carlo Simulations.
The state initial value of each target are as follows:
Target 1:[5000 300 6000-100];
Target 2:[1000 400 1,500 100];
Target 3:[1000 400 1,800 100].
For time complexity, 50 Monte Carlo simulations are completed, distributed UKF-IJPDA filter needs 2.08s,
UKF-JPDA filter needs 6.46s, UKF-IJPDA to substantially reduce time of algorithm.Therefore point based on IJPDA proposed
Cloth UKF algorithm remains the tracking accuracy as Joint Probabilistic Data Association algorithm while reducing simulation time.
Claims (3)
1. a kind of based on the distributed object tracking for improving joint probability data association, which is characterized in that including following step
It is rapid:
(1) motion model is established to each target respectively to sensor multiple-target system;
(2) data correlation filtering method is carried out to each sensor respectively, obtains respective sensor about the mesh in detection range
Target state estimation
Data correlation filtering method includes:
(2.1) tracking gate of each target is set, determines effective echo m of targetk, and initial association probability is calculated, construction determines
MatrixExpression echo, t=1,2 ... T indicates target;Determine that the element in matrix only indicates
Relationship between target and effective echo does not consider the relationship of echo and false-alarm;Each target following door is first calculated according to PDA method
Interior echo association probability
Effective echo j is derived from the probability of target t:
There is no echo to be derived from the probability of target t:
bt=λ (1-PDPG)|2πSt(k)|1/2,Wherein parameter PDIt indicates to true mesh
Target detection probability, PGIndicate that the measurement of sensor falls into the probability in tracking gate, λ indicates the density of environment clutter, St(k)=H
(k)Pt(k|k-1)HT(k)+R (k) is the new breath covariance of target,Indicate a certain measurement pair
The residual error of target;
(2.2) according to the public echo ensembles Pub for determining that matrix Ω determines moment k;
(2.3) to association probability Pt jIt is modified, j=0,1 ... mk;
By the formula of (2.2), the echo in public echo ensembles Pub is associated with two or more targets, the reality according to echo
Utilization power guarantees that the utilization rate of public echo j is no more than 1, association probability is handled as follows:
For not common echo, association probability does not influence, and does not have to amendment,
When in tracking gate without echo, there is corresponding association probabilityAssociation probability βjt(k) the constraint item of following formula need to be met
Part, makes the association probability of target and for 1,
To target association probabilityCarry out following normalized:
The association probability of echo is calculatedCorresponding state estimation covariance is obtained,
(3) respectively sensor state estimation carry out space correlation, then merge same target state estimation obtain it is final
Target state estimator value.
2. a kind of distributed object tracking based on improvement joint probability data association according to claim 1,
It is characterized in that, the step (1) specifically:
Motion model is established to each target respectively to sensor multiple-target system;Establish Nonlinear Parameter tracking system mould
Type:
For target following since the quantity of state and measurement of selection are under different coordinate systems, target following is one non-
Linear state estimation problem;W in formulak、VkThe measurement noise of system noise and respective sensor is respectively indicated, and meets item
Part:
WhereinIndicate k moment sensor i to the measurement information of target;During target following, detected at sensor
Information calculates the association probability that all echoes are derived from target by data association algorithm, then calculates corresponding state estimation.
3. a kind of distributed object tracking based on improvement joint probability data association according to claim 1,
It is characterized in that, the step (3) are as follows:
The state estimation of respective sensor carries out space correlation, and the state estimation for then merging same target obtains final target
State estimation;
Wherein, ci,lIt indicates the diffusion weight of each sensor, and meets
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610821318.5A CN106443622B (en) | 2016-09-13 | 2016-09-13 | A kind of distributed object tracking based on improvement joint probability data association |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610821318.5A CN106443622B (en) | 2016-09-13 | 2016-09-13 | A kind of distributed object tracking based on improvement joint probability data association |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106443622A CN106443622A (en) | 2017-02-22 |
CN106443622B true CN106443622B (en) | 2018-12-07 |
Family
ID=58168430
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610821318.5A Active CN106443622B (en) | 2016-09-13 | 2016-09-13 | A kind of distributed object tracking based on improvement joint probability data association |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106443622B (en) |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106291533B (en) * | 2016-07-27 | 2018-10-16 | 电子科技大学 | A kind of distributed multi-sensor fusion method based on AMD |
CN107271991B (en) * | 2017-05-25 | 2019-05-24 | 北京环境特性研究所 | A kind of optical electrical sensor target correlating method based on state estimation |
CN107358121B (en) * | 2017-07-12 | 2018-10-02 | 张�诚 | A kind of data fusion method and device of desensitization data set |
CN107688179B (en) * | 2017-08-07 | 2020-01-03 | 上海无线电设备研究所 | Comprehensive probability data interconnection method based on Doppler information assistance |
CN109597104A (en) * | 2018-10-22 | 2019-04-09 | 中国科学院光电研究院 | A method of target following is positioned using multi-model multisystem |
CN110221307B (en) * | 2019-05-28 | 2022-12-13 | 哈尔滨工程大学 | Multi-passive sonar non-cooperative multi-target line spectrum information fusion method |
CN110208767B (en) * | 2019-06-05 | 2022-12-13 | 哈尔滨工程大学 | Radar target rapid detection method based on fitting correlation coefficient |
CN111684457B (en) * | 2019-06-27 | 2024-05-03 | 深圳市大疆创新科技有限公司 | State detection method and device and movable platform |
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 |
CN111007454B (en) * | 2019-10-28 | 2022-07-01 | 重庆邮电大学 | Extended target tracking method based on cooperative target information |
CN112654979B (en) * | 2020-04-29 | 2021-12-17 | 华为技术有限公司 | Data association method and device |
CN111767639B (en) * | 2020-05-25 | 2022-12-13 | 西北工业大学 | Multi-sensor track association method |
CN112146648B (en) * | 2020-09-23 | 2022-08-19 | 河北工业大学 | Multi-target tracking method based on multi-sensor data fusion |
CN112748735B (en) * | 2020-12-18 | 2022-12-27 | 重庆邮电大学 | Extended target tracking method introducing color features |
CN113191427B (en) * | 2021-04-29 | 2022-08-23 | 无锡物联网创新中心有限公司 | Multi-target vehicle tracking method and related device |
CN113376626A (en) * | 2021-06-23 | 2021-09-10 | 西安电子科技大学 | High maneuvering target tracking method based on IMMPDA algorithm |
CN113486300A (en) * | 2021-07-02 | 2021-10-08 | 南通大学 | Unmanned vehicle multi-target tracking method |
CN113484866B (en) * | 2021-07-05 | 2022-04-29 | 哈尔滨工程大学 | Multi-target detection tracking method based on passive sonar azimuth history map |
CN113532422B (en) * | 2021-07-12 | 2022-06-21 | 哈尔滨工程大学 | Multi-sensor track fusion method based on distance map and data cleaning |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004219299A (en) * | 2003-01-16 | 2004-08-05 | Mitsubishi Electric Corp | Parallel multiple target tracking system |
CN101783020A (en) * | 2010-03-04 | 2010-07-21 | 湖南大学 | Video multi-target fast tracking method based on joint probability data association |
CN102521612A (en) * | 2011-12-16 | 2012-06-27 | 东华大学 | Multiple video object active tracking method based cooperative correlation particle filtering |
CN105137418A (en) * | 2015-07-28 | 2015-12-09 | 中国人民解放军海军航空工程学院 | Multi-object tracking and data interconnection method based on whole neighborhood fuzzy clustering |
-
2016
- 2016-09-13 CN CN201610821318.5A patent/CN106443622B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004219299A (en) * | 2003-01-16 | 2004-08-05 | Mitsubishi Electric Corp | Parallel multiple target tracking system |
CN101783020A (en) * | 2010-03-04 | 2010-07-21 | 湖南大学 | Video multi-target fast tracking method based on joint probability data association |
CN102521612A (en) * | 2011-12-16 | 2012-06-27 | 东华大学 | Multiple video object active tracking method based cooperative correlation particle filtering |
CN105137418A (en) * | 2015-07-28 | 2015-12-09 | 中国人民解放军海军航空工程学院 | Multi-object tracking and data interconnection method based on whole neighborhood fuzzy clustering |
Non-Patent Citations (3)
Title |
---|
一种改进的联合概率数据关联算法的研究;唐冬丽 等;《弹箭与制导学报》;20101231;第30卷(第6期);第18-20页 * |
吴佳芯.多目标跟踪的数据关联算法研究.《中国优秀硕士学位论文全文数据库 信息科技辑》.2013, * |
多目标跟踪的数据关联算法研究;冯洋;《中国优秀硕士学位论文全文数据库 信息科技辑》;20080115;第I136-462页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106443622A (en) | 2017-02-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106443622B (en) | A kind of distributed object tracking based on improvement joint probability data association | |
CN103729859B (en) | A kind of probability nearest neighbor domain multi-object tracking method based on fuzzy clustering | |
CN112613532B (en) | Moving target tracking method based on radar and cyclic neural network complement infrared fusion | |
CN107066806B (en) | Data Association and device | |
CN106980114A (en) | Target Track of Passive Radar method | |
CN106646450A (en) | Radar track anti-error correlation method based on distance stepped clustering | |
CN106291533A (en) | A kind of distributed multi-sensor blending algorithm based on AMD | |
CN105137418B (en) | Multiple target tracking and data interconnection method based on complete adjacent fuzzy clustering | |
CN104715154B (en) | Core K average Data Associations based on KMDL criterion criterions | |
CN109858526A (en) | Sensor-based multi-target track fusion method in a kind of target following | |
CN111711432B (en) | Target tracking algorithm based on UKF and PF hybrid filtering | |
Ma et al. | Target tracking system for multi-sensor data fusion | |
CN103247057A (en) | Road target multi-hypothesis tracking algorithm under target-echo-road network data association | |
CN111830501B (en) | HRRP history feature assisted signal fuzzy data association method and system | |
CN109934849A (en) | Online multi-object tracking method based on track metric learning | |
CN109115228A (en) | A kind of object localization method based on weighted least-squares volume Kalman filtering | |
CN117169818B (en) | Radar waveform design method for sea surface maneuvering target tracking | |
Li et al. | An improved observation model for Monte-Carlo localization integrated with reliable reflector prediction | |
CN109031278A (en) | It is a kind of based on amplitude to extra large Radar Multi Target fusion tracking method | |
Li et al. | A track initiation method for the underwater target tracking environment | |
CN104237880B (en) | Structure changes Joint Probabilistic Data Association formation target tracking method | |
Wang et al. | Research on multi-maneuvering target tracking JPDA algorithm | |
Zhang et al. | A Fusion Localization Algorithm Combining MCL with EKF | |
CN114088086B (en) | Multi-target robust positioning method for resisting measurement wild value interference | |
Hai-peng et al. | A group target track correlation algorithm based on systematic error estimation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |