CN105894535B - A kind of vortex method for automatic tracking based on Bayes - Google Patents
A kind of vortex method for automatic tracking based on Bayes Download PDFInfo
- Publication number
- CN105894535B CN105894535B CN201610191428.8A CN201610191428A CN105894535B CN 105894535 B CN105894535 B CN 105894535B CN 201610191428 A CN201610191428 A CN 201610191428A CN 105894535 B CN105894535 B CN 105894535B
- Authority
- CN
- China
- Prior art keywords
- vortex
- tracker
- tracking
- observation
- prediction
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Landscapes
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of vortex method for automatic tracking based on Bayes, it is modeled using motion process of the Kalman filter based on Bayes to vortex, tracking and matching is carried out by predicted position and observed result, and solves the problems, such as the matching conflict of multiple target vortex tracking using Hungarian optimization algorithm.The present invention overcomes the complex match problems in multiple target vortex tracing process, improve the accuracy of tracking result.
Description
Technical field
The present invention relates to a kind of vortex method for automatic tracking based on Bayes belongs to ocean eddy automatic Extraction Algorithm neck
Domain.
Background technique
Mesoscale eddy (being referred to as vortexed) is widely distributed and very important a kind of dynamic phenomenon in ocean, it can not only be passed
It passs heat, propagate nutriment, while being also the major embodiment of ocean kinetic energy.The development of Modern remote observation technology and application are
The differentiation Evolution for studying ocean eddy provides observational data abundant for a long time.From the observational data of these magnanimity
The evolutionary process for extracting vortex is excavated, is the basic premise for studying eddying motion Evolution.
Currently, there are mainly three types of the method for tracing of vortex evolutionary process: 1) range search method, mainly with current tracking vortex
Central point is reference settings search radius, fall into vortex in radius in the search of next moment, and selected distance is nearest
It is vortexed as evolution subsequent (Chelton et al.2011);2) similitude is tracked, and on the basis of range search method, is added
The measurement of similarity degree between vortex develops subsequent (Chaigneau et al.2008) with most like vortex or so;3) face
Product overlay method develops front and back spatial shape with the presence or absence of overlapping according to being vortexed, developed to determine whether belonging to a vortex
Journey (Henson et al.2008).
These existing methods, algorithm idea is simple, is widely used, but there are two o'clock deficiencies: moving first is that having ignored vortex
Dynamic essence, either half path search or area overlapping, if being with reference to come when tracking next with the position being currently vortexed
It carves, substantially assumes that the next moment that is vortexed remains in original place and do not move, and this is not in accordance with facts;Second is that working as multiple whirlpools
Rotation is when being closely located to, the tracking and matching problem of complexity easy to form, such as two vortexs next moment track three it is potential
Subsequent, which kind of matching is just best suitable for vortex's motion feature actually, and there are three types of all do not answer accordingly in method now for this problem
To method.
Summary of the invention
Present invention solves the technical problem that: a kind of vortex method for automatic tracking based on Bayes is provided, using Kalman
Filtering models eddying motion process, solves the problems, such as the prediction of vortex subsequent time position, while using Hungarian
Optimization matching algorithm solves the problems, such as matching conflict when multiple vortex target trackings, improves the accuracy of tracking result.
Technical solution of the present invention: a kind of vortex method for automatic tracking based on Bayes, first building are based on Bayes
Kalman filter (Bar-Shalom et al.2001) motion process of vortex is modeled, from probability predict be vortexed
The position and effective error scope that next moment most probable occurs, then tracking and matching is carried out with the result of observation, if there is
Multi-target tracking matching, then search for each vortex mesh using Hungarian optimization matching algorithm (J.Munkres 1957)
Target is most preferably subsequent.
Specific step is as follows:
Step 1, initialization, establish an empty vortex tracker set, one be vortexed observation value set and one it is empty
Tracking result set.Each vortex tracker will be after the preamble for recording a vortex evolutionary process during automatic tracing
After relationship.The observation value set that is vortexed will record the eddy information (position, attribute etc.) that each moment observes.Tracking result set
The vortex tracker that storage record life process tracking is completed.To each vortex that the first moment observes, building be vortexed with
Track device, records the location information of vortex, and then the Kalman filter for creating default parameters adds tracker for tracking prediction
It is added in vortex tracker set;
Step 2, starting tracking, since next moment, temporally frame is recycled, if being recycled to last frame,
Tracking stops, the vortex tracker that tracking result set is collected, and has recorded the Life Evolution procedural information that tracking is vortexed;Otherwise,
Automatic tracing is carried out, step 3-4 is executed;
Step 3, all vortex observations for obtaining current time frame are added in observation value set;
Step 4, data correlation, each of traversal tracker set vortex tracker, first carry out Kalman filter
Tracking prediction, then the vortex of the vortex position of prediction and error range and current time is observed into all vortexs in value set,
It is associated matching:
If ● the vortex being not matched in the prediction error range of vortex tracker in observation value set, by the tracking
Device takes out from set, is added in the tracking result set of step 1 foundation, shows that the Life Evolution process of the vortex has chased after
Track terminates;
● if one and only one matched vortex observation in the prediction error range of vortex tracker is seen from being vortexed
The vortex is taken out in measured value set, as the subsequent vortex that tracker traces into, and using Kalman filter to the vortex
Observation position is modified, and obtains filtered vortex position and error range, and the process trace after carrying out of taking this as the standard;
If ● there are multiple matched vortex observations in the prediction error range of vortex tracker, use first
The bipartite graph (bipartite graph) that Hungarian algorithm constitutes the tracker and vortex observation carries out optimization
Match, for being matched to the vortex tracker of observation, which is taken out from observation value set, is traced into as tracker
Subsequent vortex, and the observation position of the vortex is modified using Kalman filter, obtains filtered vortex position
And error range, and the process trace after carrying out of taking this as the standard;
If ● there is the vortex of non-matched jamming device in observation value set, creates a vortex tracker, be added to whirlpool
It revolves in tracker set.
The advantages of the present invention over the prior art are that: the Kalman filter based on Bayes is used, more accurately mould
The quasi- motion process being vortexed;The complex match conflict problem that multi-target tracking is solved using Hungarian algorithm, improves tracking
Accuracy.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention.
Specific embodiment
As shown in Figure 1, specific implementation step of the invention is as follows:
Step 1, initialization, establish an empty vortex tracker set (trackerList), a vortex observation collection
Close (observeList) and an empty tracking result set (completeList).Each vortex tracker will be automatic
By recording vortex ID in Ids attribute during tracking, the subsequent relationship of preamble of vortex life process is stored.Be vortexed observation
Value set will record the vortex position that each moment observes.Tracking result set completes storage record life process tracking
Vortex tracker.To each vortex that the first moment observes, vortex tracker is constructed, the position of vortex is recorded in Obs attribute
Vector is set, and then tracker is added to vortex tracker collection for tracking prediction by the Kalman filter for creating default parameters
In conjunction.
Tracker, the data structure of observation are as follows, tracking result set will collect, storage tracked the vortex of completion with
Track device:
Tracker
Observation
OID | Obs |
Vortex ID | Position detection value vector |
Step 2, starting tracking, since next moment, temporally frame is recycled, if being recycled to last frame,
Tracking stops, the vortex tracker that tracking result set is collected, and has recorded the Life Evolution procedural information that tracking is vortexed;Otherwise,
Automatic tracing is carried out, step 3-4 is executed;
Step 3, all vortex observations for obtaining current time frame are added in observation value set observeList;
Step 4, data correlation.Each of tracker set trackerList vortex tracker is traversed, is first carried out
The tracking prediction of Kalman filter belongs to Pre and Cov that the vortex position of prediction and covariance matrix are respectively stored into tracker
Property in, and calculate and mahalanobis (geneva) distance of vortex observation position.Mahalanobis distance calculation formula is as follows:
Wherein, X indicates that predicted value vector, Y indicate that observation vector, Σ indicate the covariance matrix of Kalman filter.
Since mahalanobis distance meets chi square distribution, so when the predicted value vector of vortex tracker and the mahalanobis distance of observation vector are big
When 0.05 level of signifiance, it is believed that observation is outside the error range of prediction, and when tracking and matching should foreclose, conversely, whirlpool
Observation is revolved in the error range of prediction, carries out tracking and matching by following several situations:
If ● the vortex being not matched in the prediction error range of vortex tracker in observation value set, by the tracking
Device takes out from trackerList, is added in the tracking result set completeList of step 1 foundation, shows the vortex
Life Evolution process tracked and terminated;
● if one and only one matched vortex observation in the prediction error range of vortex tracker is seen from being vortexed
The vortex is taken out in measured value set observeList, as the subsequent vortex that tracker traces into, with the spatial position of the vortex
The Obs attribute of tracker is updated, meanwhile, it is modified with position of the Kalman filter in tracker to vortex, with filtering
Vortex position and error matrix afterwards updates Upd the and Cov attribute of tracker, and the process trace after carrying out of taking this as the standard;
● if there are multiple matched vortex observations in the prediction error range of vortex tracker, first using figure time
It goes through algorithm and extracts tracker and observation in bipartite graph, the mahalanobis distance calculated between tracker and observation obtains distance
Then Cost matrix carries out optimization matching to the distance costs matrix using Hungarian algorithm.For being matched to observation
Vortex tracker, from be vortexed observation value set observeList in take out the vortex, the subsequent whirlpool traced into as tracker
Rotation updates the Obs attribute of tracker with the spatial position of the vortex, and with the Kalman filter in tracker to the position of vortex
Set and be modified, Upd the and Cov attribute of tracker is updated with filtered vortex position and error matrix, and take this as the standard into
Process trace after row;
If ● there is the vortex of non-matched jamming device in observation value set, creates a vortex tracker, be added to whirlpool
It revolves in tracker set.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this neighborhood
For art personnel, the present invention can change and change.All within the spirits and principles of the present invention, it is made it is any modification, etc.
With replacement, improvement etc., should be included within scope of the presently claimed invention.
Claims (1)
1. a kind of vortex method for automatic tracking based on Bayes, it is characterised in that steps are as follows:
Step 1, initialization establish an empty vortex tracker set, a vortex observation value set and an empty tracking
Results set;Each vortex tracker will record the preamble subsequent pass of a vortex evolutionary process during automatic tracing
System;The observation value set that is vortexed will record the eddy information that each moment observes, tracking result set will storage record life mistake
The vortex tracker that journey tracking is completed;To each vortex that the first moment observes, vortex tracker is constructed, the position of vortex is recorded
Confidence breath, and then tracker is added to vortex tracker collection for tracking prediction by the Kalman filter for creating default parameters
In conjunction;
Step 2, starting tracking, since next moment, temporally frame is recycled, if being recycled to last frame, tracking
Stop, the vortex tracker that tracking result set is collected, has recorded the Life Evolution procedural information that tracking is vortexed;Otherwise, it carries out
Automatic tracing executes step 3-4, until end time ability end loop, stops tracking;
Step 3, all vortex observations for obtaining current time frame are added in observation value set;
Step 4, data correlation, each of traversal tracker set vortex tracker, first carry out the tracking of Kalman filter
Prediction, then by all vortexs in the vortex observation value set at the vortex position of prediction and error range and current time, carry out
Association matching;
Data correlation in the step 4, each of traversal tracker set vortex tracker, first carries out Kalman filter
Tracking prediction, then the vortex of the vortex position of prediction and error range and current time is observed into all vortexs in value set,
It is associated matched process:
(41) if the vortex in observation value set is not matched in the prediction error range of vortex tracker, by the tracker
It is taken out from set, is added in the tracking result set of step 1 foundation, shows that the Life Evolution process of the vortex has been tracked
Terminate;
(42) if one and only one interior matched vortex observation of the prediction error range of vortex tracker, is observed from being vortexed
The vortex is taken out in value set, as the subsequent vortex that tracker traces into, and the sight using Kalman filter to the vortex
Location, which is set, to be modified, and filtered vortex position and error range are obtained, and the process trace after carrying out of taking this as the standard;
(43) if there are multiple matched vortex observations in the prediction error range of vortex tracker, in order to guarantee tracking result
Accuracy, the bipartite graph (bipartite of multiple observations Yu multiple trackers is constructed using Hungarian algorithm
Graph), and in carrying out one-to-one matched solution procedure, multi-track is determined with global distance and minimum optimal objective
The Optimum Matching scheme of device and more observations, for being matched to the vortex tracker of observation, by the observation from observation collection
It takes out in conjunction, is carried out as the subsequent vortex that tracker traces into, and using observation position of the Kalman filter to the vortex
Amendment, obtains filtered vortex position and error range, and the process trace after carrying out of taking this as the standard;
(44) if there is the vortex of non-matched jamming device in observation value set, a vortex tracker is created, vortex is added to
In tracker set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610191428.8A CN105894535B (en) | 2016-03-30 | 2016-03-30 | A kind of vortex method for automatic tracking based on Bayes |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610191428.8A CN105894535B (en) | 2016-03-30 | 2016-03-30 | A kind of vortex method for automatic tracking based on Bayes |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105894535A CN105894535A (en) | 2016-08-24 |
CN105894535B true CN105894535B (en) | 2019-05-31 |
Family
ID=57014673
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610191428.8A Active CN105894535B (en) | 2016-03-30 | 2016-03-30 | A kind of vortex method for automatic tracking based on Bayes |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105894535B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107292911B (en) * | 2017-05-23 | 2021-03-30 | 南京邮电大学 | Multi-target tracking method based on multi-model fusion and data association |
CN107657148B (en) * | 2017-11-09 | 2021-09-28 | 中国海洋大学 | Breakpoint continuous chasing method based on long-time span global mesoscale vortex |
CN111695299B (en) * | 2020-06-04 | 2022-12-13 | 哈尔滨工程大学 | Mesoscale vortex trajectory prediction method |
CN112767711B (en) * | 2021-01-27 | 2022-05-27 | 湖南优美科技发展有限公司 | Multi-class multi-scale multi-target snapshot method and system |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102148921A (en) * | 2011-05-04 | 2011-08-10 | 中国科学院自动化研究所 | Multi-target tracking method based on dynamic group division |
CN102289594A (en) * | 2011-08-19 | 2011-12-21 | 中国科学院地理科学与资源研究所 | Algorithm for automatically identifying and reconstructing process of mesoscale ocean eddy |
CN102737386A (en) * | 2012-05-15 | 2012-10-17 | 北京硅盾安全技术有限公司 | Moving target anti-fusion shielding tracking algorithm |
CN102915545A (en) * | 2012-09-20 | 2013-02-06 | 华东师范大学 | OpenCV(open source computer vision library)-based video target tracking algorithm |
CN103281476A (en) * | 2013-04-22 | 2013-09-04 | 中山大学 | Television image moving target-based automatic tracking method |
CN103399905A (en) * | 2013-07-29 | 2013-11-20 | 中国科学院地理科学与资源研究所 | Method of information extraction for ocean eddy evolution process |
US8705793B2 (en) * | 2009-10-08 | 2014-04-22 | University Of Southern California | Object tracking by hierarchical association of detection responses |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7626535B2 (en) * | 2006-11-09 | 2009-12-01 | Raytheon Company | Track quality based multi-target tracker |
-
2016
- 2016-03-30 CN CN201610191428.8A patent/CN105894535B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8705793B2 (en) * | 2009-10-08 | 2014-04-22 | University Of Southern California | Object tracking by hierarchical association of detection responses |
CN102148921A (en) * | 2011-05-04 | 2011-08-10 | 中国科学院自动化研究所 | Multi-target tracking method based on dynamic group division |
CN102289594A (en) * | 2011-08-19 | 2011-12-21 | 中国科学院地理科学与资源研究所 | Algorithm for automatically identifying and reconstructing process of mesoscale ocean eddy |
CN102737386A (en) * | 2012-05-15 | 2012-10-17 | 北京硅盾安全技术有限公司 | Moving target anti-fusion shielding tracking algorithm |
CN102915545A (en) * | 2012-09-20 | 2013-02-06 | 华东师范大学 | OpenCV(open source computer vision library)-based video target tracking algorithm |
CN103281476A (en) * | 2013-04-22 | 2013-09-04 | 中山大学 | Television image moving target-based automatic tracking method |
CN103399905A (en) * | 2013-07-29 | 2013-11-20 | 中国科学院地理科学与资源研究所 | Method of information extraction for ocean eddy evolution process |
Non-Patent Citations (2)
Title |
---|
Multiple Moving Objects Tracking under Complex Scenes;Qin Wan 等;《Proceedings of the 6th World Congress on Intelligent Control》;20060623;第9871-9875页 * |
郝菲.智能监控系统中运动目标检测跟踪的研究.《中国优秀硕士学位论文全文数据库•信息科技辑》.2010,(第2期),第33-49页. * |
Also Published As
Publication number | Publication date |
---|---|
CN105894535A (en) | 2016-08-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105894535B (en) | A kind of vortex method for automatic tracking based on Bayes | |
Andriyenko et al. | An analytical formulation of global occlusion reasoning for multi-target tracking | |
CN106772246B (en) | Unmanned plane real-time detection and positioning system and method based on acoustic array | |
Badino et al. | Visual topometric localization | |
Shen et al. | Fast online tracking with detection refinement | |
CN106327526A (en) | Image object tracking method and image object tracking system | |
CN103176164B (en) | Based on the Multi-Target Passive Tracking method of wireless sensor network | |
CN106355604A (en) | Target image tracking method and system | |
CN111708021B (en) | Personnel tracking and identifying algorithm based on millimeter wave radar | |
CN108182413A (en) | A kind of mine movable object detecting and tracking recognition methods | |
CN104835178A (en) | Low SNR(Signal to Noise Ratio) motion small target tracking and identification method | |
CN109557532B (en) | Tracking method before detection based on three-dimensional Hough transform and radar target detection system | |
CN106054169A (en) | Multi-station radar signal fusion detection method based on tracking information | |
CN106291498B (en) | A kind of detecting and tracking combined optimization method based on particle filter | |
CN108734725A (en) | Probabilistic contractor couple based on Gaussian process extends method for tracking target | |
CN110532921A (en) | The more Bernoulli Jacob's video multi-target trackings of broad sense label are detected based on SSD | |
Zhao et al. | A novel measurement data classification algorithm based on SVM for tracking closely spaced targets | |
Chen et al. | Motion propagation detection association for multi-target tracking in wide area aerial surveillance | |
CN112651994A (en) | Ground multi-target tracking method | |
Bashar et al. | Multiple object tracking in recent times: A literature review | |
Hao et al. | Asynchronous information fusion in intelligent driving systems for target tracking using cameras and radars | |
Ma et al. | Radiation intensity Gaussian mixture PHD filter for close target tracking | |
CN103413138B (en) | Point target detecting method in a kind of infrared image sequence | |
Liu et al. | Multi-sensor tracking with non-overlapping field for the GLMB filter | |
Chen et al. | Efficient online tracking-by-detection with Kalman filter |
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 |