CN108090922A - Intelligent Target pursuit path recording method - Google Patents
Intelligent Target pursuit path recording method Download PDFInfo
- Publication number
- CN108090922A CN108090922A CN201611025664.9A CN201611025664A CN108090922A CN 108090922 A CN108090922 A CN 108090922A CN 201611025664 A CN201611025664 A CN 201611025664A CN 108090922 A CN108090922 A CN 108090922A
- Authority
- CN
- China
- Prior art keywords
- target
- camera
- track
- image
- sample
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- 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
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30241—Trajectory
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The present invention relates to intelligent Target pursuit path recording methods, and target image is gathered using multiple cameras, by estimating away from the three-dimensional track of target movement is obtained, comprise the following steps more:When each camera detects the target in acquisition image to track target, using KCF track algorithms to tracking target into line trace;The position of target obtains the three-dimensional coordinate of target in the image gathered according to each camera;And it records three-dimensional coordinate and is shown.The present invention tracks target using improved M KCF high speeds track algorithm, and high certainty of measurement, speed is fast, suitable for the situation that target translational speed is very fast or system real time is more demanding.
Description
Technical field
The present invention relates to computer tracking fields, track target particular by track algorithm, then pass through principle of parallax
The three-dimensional coordinate of target is calculated, eventually forms a kind of intelligent Target pursuit path recording method of track.
Background technology
The tracking of target is an important research field of computer vision.With the development of science and technology, target following and
Target trajectory be recorded in traffic monitoring, pedestrian's flow, astronomical observation, automatic Pilot, aircraft research and development etc. fields have it is very practical
Value.For target following, largely scholar has done many work both at home and abroad.Common target tracking algorism has been almost now
It can achieve the purpose that real-time tracking.But in some fields, such as the research and development field of aircraft or real-time to target following
In the more demanding field of property.Since target velocity is very fast or requirement of real-time is higher, traditional tracking can not reach
To the purpose of real-time tracking.
The content of the invention
It cannot be with tracking the target that quickly move or system real-time itself is not good etc. enough asks for previous tracking system
Topic, present invention proposition track target using improved high speed track algorithm, are finally reached the purpose of record target trajectory.
Present invention technical solution used for the above purpose is:Intelligent Target pursuit path recording method uses
Multiple cameras gather target image, by estimating away from the three-dimensional track of target movement is obtained, comprise the following steps more:
When each camera detects the target in acquisition image to track target, using KCF track algorithms to tracking mesh
It marks into line trace;
The position of target obtains the three-dimensional coordinate of target in the image gathered according to each camera;
And it records three-dimensional coordinate and is shown.
The KCF track algorithms comprise the following steps:
1) use tracking target in all camera current acquired images as positive sample, to positive sample cyclic shift it
After obtain negative sample;
2) use positive sample, negative sample update KCF track algorithms obtain the parameter w of SVM classifier in KCF track algorithms with
And constant w0;
3) each camera inputs next frame acquisition image respectively, and the feature x of next frame acquisition image is substituted into letter respectively
Number f (x)=wTx+w0;X be image feature, w0For the constant for plane of classifying;
Judge whether f (x) maximums of each camera are more than 0;
If when the maximum of f (x) in preceding camera is more than the corresponding feature x position of 0, f (x) maximums for current camera shooting
Target location in head next frame acquisition image;Otherwise, then target is lost, and stops tracking;
When all cameras judge to finish, return to step 1), until target disappears or stops manually.
It is described to be specially to obtaining negative sample after positive sample cyclic shift:
Positive sample with the circular matrix being made of unit matrix is multiplied, forms negative sample;
C (x) is the circular matrix of a n × n, is obtained by the cyclic shift of the vector x of a 1 × n:
The w is obtained by following steps:
X is brought into following formulas to obtain
W=(XXT+λI)-1XTy
Wherein, X is gained matrix after circular matrix progress discrete Fourier transform;Y expression sample labels, positive sample 1,
Negative sample is -1;λ is parameter;I is unit matrix.
The three-dimensional coordinate that the position of target obtains target in the image of each camera acquisition of the basis is specially:
The target location in every two field picture in each camera obtains target and the distance of setting origin in every frame;Root
According to the target location in the distance of every frame and the every two field picture of each camera, three-dimensional coordinate of the target per frame is obtained.
The invention has the advantages that and advantage:
1st, using Industrial PC as host computer, measuring system forms simple in structure the present invention, reliability is high, it is at low cost,
Performance is high.
2nd, inventive algorithm supports minimum 2 or more video cameras, and the more precision of video acquisition unit quantity are higher, quantity is few
Then dispose simplicity.
3rd, the present invention tracks target using improved M-KCF high speeds track algorithm, and high certainty of measurement, speed is fast, is suitable for
The situation that target translational speed is very fast or system real time is more demanding.
4th, the present invention realizes that data processing calculates by computer-controlled program, and computational accuracy is high, passes through numerical monitor list
Member is accurate, intuitively shows test result.
5th, the present invention is using Industrial PC, and system composition is simple in structure, and reliability is high, at low cost;Target track in hundred meters
Mark record quality is quite high, and speed is fast, greatly reduces the input of observation person.
Description of the drawings
Fig. 1 is whole system structural framing figure of the present invention;
Fig. 2 is recognition and tracking system program flow chart of the present invention;
Fig. 3 calculates record system program flow chart for coordinate of the present invention;
Fig. 4 is the data management module block diagram of the present invention.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and embodiments.
The present invention discloses intelligent Target track following record system.The system includes host computer, double (more) mesh video acquisitions
Unit and display module.The host computer includes intelligent-tracking, three-dimensional track logging modle.Host computer receives video acquisition list
The digital signal of member, the host computer are remembered according to the digital signal received by intelligent-tracking, three-dimensional track logging modle
The running orbit of target is recorded, wherein, intelligent tracking system is tracked by using high-speed target under target identification and complex background
Algorithm identifies, calculates and records the running orbit of target, and the running orbit recorded is output to display module afterwards shows;
The video acquisition unit is double (more) mesh cameras, gathers the object run track video of different angle, then inputs upper
Machine;The display module generates the Three-dimensional Display of object run track by receiving the object run track that host computer inputs
Figure.The system leads to the tracking of trace routine realization target, the calculating of track and record, the three-dimensional rail of last display target operation
Mark.
A kind of intelligent Target pursuit path records system, and hardware components include host computer, double (more) mesh video acquisition units
And display module:
Host computer receives the digital signal of video acquisition unit, and the host computer will carry out intelligence to the digital signal received
Can identification, tracking, trajectory calculation and record, finally export to display module and show;
The video acquisition unit signal output part is connected with host computer;
The display module receives the coordinate of host computer output object run track, and is output to display portion.
Storage control program in host computer, including target identification module, tracking module, the three-dimensional seat in object run track
Mark computing module, Grid Track memory module.
The system performs following measuring process by pc control procedure:
Step 1) starts control program initialization module, and the basic parameter of whole system is set, is set by above-mentioned parameter
Put the running of final control whole system.
Step 2) starts target identification module, waits target into double (more) mesh video acquisition units.Target is waited to enter
After double (more) mesh video acquisition units, target identification module carrys out judgment object.Whether it is to need the target tracked.
Step 3) start-up trace module, when step 2) is when being determined as true.After tracking module starts, mesh is run to always
Mark disappears or manual operation stops.
Step 4) starts the three-dimensional coordinate calculating of object run track, logging modle, when step 3) stops.Calculate target
Running orbit three-dimensional coordinate is simultaneously recorded in host computer hard disk.
Step 5) starts the object run track three-dimensional coordinate that display module receives host computer output, ultimately produces visualization
Three-dimensional track is shown to user.
Step 6) log-on data management module, the data management module include test information, measured object information, operative position
It puts, tables of data, store, recall and print.
Program initialization module is controlled to carry out the setting of basic parameter to whole system, including camera quantity and parameter
Setting needs the target signature identified, target sizes, the threshold value including triggering tracking, the parameter of track algorithm, target trajectory to sit
Mark parameter of coordinate fitting algorithm etc. in the parameter calculated, display module.
The purpose of target identification module is target of the capture into video acquisition unit.When target enters video acquisition unit
During acquisition range, start the recognizer in identification module to differentiate and judge whether into object be to need the target tracked.
Target tracking module is that the position of target in video acquisition unit is determined using algorithm, and target tracking module receives
Video acquisition unit is input to the video data in host computer, according to the target position of the image of each frame of data and previous frame
Put to judge the position of target in this frame of video.And the location information of current goal is recorded.
Three-dimensional coordinate calculates, logging modle is in the video that the target tracked according to track algorithm is shot in different angle
The target three-dimensional coordinate that the data such as the world coordinates of position and each camera are calculated.The three-dimensional calculated for every frame
Coordinate, system are all recorded, and then serve data to display module.
Display module is set including target trajectory fitting display area, displacement of targets track numerical value display area, systematic parameter
It puts and tables of data area.
Displacement of targets track fitting is shown:It shows the target trajectory three-dimensional coordinate point measured and fits the target come
Running orbit curve.
System parameter setting:Measurement method, parameter are set.
Displacement of targets track numerical value is shown:The display of measurement data preserves.
Data management module, wherein metrical information:Including tester and testing time;Measured target information includes measured object
Number, species;Operating position realizes the path of selection data storage file and the name of file;Tables of data measurement data is shown in rule
In model data sheet;The preservation of measurement data is realized in storage;Recall the inquiry for realizing measurement data;Data sheet is realized in printing
Printing.As shown in Figure 4.
Energy target following track record system, hardware components include host computer, double (more) mesh video acquisition units and display
Module, the host computer receive the digital signal of video acquisition unit, and the host computer will carry out the digital signal received
Intelligent recognition, tracking, trajectory calculation simultaneously record, finally export to display module and show;The video acquisition unit signal output
End is connected with host computer;The display module receives the coordinate of host computer output object run track, and is output to display unit
Point.
The MV-GE30GC Mai Dewei that video acquisition unit regards company using Mai Dewei regard kilomega network industrial camera;It is described to regard
Frequency collecting unit valid pixel is 752H × 480V (360,000), and default resolution ratio reaches 640X480@ROI@76FPS, dynamic range
Reach 55dB, exposure time range is 0.0285 to 54.72 milliseconds, support several operation systems and C/C++, VB6, BCB,
A variety of programming softwares such as VB.net, Delphi6, C#, Labview.Camera is placed on holder, and collecting unit is two camera shootings
In the case of head, two cameras are placed in parallel, and are directed at same direction;Three or during above camera camera it
Between be mutually an angle of 90 degrees as far as possible, and be directed at same direction.
Intelligent Target pursuit path records system, which performs following measuring process by pc control procedure:
Step 1) starts control program initialization module, and the basic parameter of whole system is set, is set by above-mentioned parameter
Put the running of final control whole system.
Step 2) starts target identification module, waits target into double (more) mesh video acquisition units.Target is waited to enter
After within the acquisition range of double (more) mesh video acquisition units, target identification module receives video acquisition by Target Recognition Algorithms
The video that unit is passed to comes whether judgment object is to need the target tracked.In the case that target is needs the target tracked,
Target information is passed to and starts target tracking module.
Step 3) start-up trace module.When step 2) is when being determined as true, start-up trace module receives video acquisition unit
The image of input calculates in present frame the position of target and temporary according to the characteristic of the position of target in previous frame and target
Record these information.After tracking module starts, target is run to always and disappears or is manually operated in video acquisition unit
Stop.
Step 4) starts the three-dimensional coordinate calculating of object run track, logging modle, when step 3) stops.Three-dimensional coordinate meter
It calculates module and object run track three-dimensional coordinate is calculated according to the target position information recorded in target tracking module.Logging modle
Then object run track three-dimensional coordinate computing module is come out and is recorded in host computer hard disk.
Step 5) starts the object run track three-dimensional coordinate that display module receives host computer output, ultimately produces visualization
Three-dimensional track is shown to user.
Step 6) log-on data management module, the data management module include test information, measured object information, operative position
It puts, tables of data, store, recall and print.
The control program initialization module carries out whole system the setting of basic parameter, including camera quantity and ginseng
Several setting needs the target signature identified, target sizes, the threshold value including triggering tracking, the parameter of track algorithm, target track
Parameter that mark coordinate calculates, parameter of coordinate fitting algorithm etc. in display module.
The purpose of the target identification module is accurately to capture into the object in video acquisition unit acquisition range to be
The no target tracked for needs.In order to reach this target, target identification module establishes video acquisition using mixed Gauss model
The background of unit acquisition.In the case that no target enters video acquisition unit, the picture captured by video acquisition unit is made
For background data.The each two field picture and background image subtraction gathered after background picture is obtained with video acquisition unit,
The place that the pixel of image has significant change is exactly the object space into video acquisition unit.So target identification module is with regard to energy
Access the object space into video acquisition unit acquisition range.
Afterwards using trained SVM classifier come judge into video acquisition unit acquisition range object whether
To need the target tracked.Svm classifier is to find classification plane f (x)=wTx+w0Come maximize in two classes apart from closest approach it
Between distance.w、w0, x difference presentation class plane equation coefficient, constant and variable.
To maximum to classification plan range apart from closest approach in two classes, the distance for being set to classifying face is r.Like this
Then have:
xpRepresent projections of the point x to classification plane, r represents x to the distance of classification plane.It is obtained after bringing into:
It is apart from the distance between closest approach in two classes:
f(x+)、f(x-) value that positive negative sample should obtain afterwards by computing is represented respectively;x+x-Positive and negative sample is represented respectively
This feature vector;yiRepresent the label of sample, positive sample is that 1 negative sample represents sample sequence number for -1, i;W is classification plane system
Number, the hyperplane that can like this maximize interval are exactly:
It is equivalent to:
The optimization problem is a quadratic programming problem (object function is quadratic function, is constrained to linear restriction), this is
The optimization problem of one standard can be solved the dual form that is written as of the problem equivalent using Lagrangian.So
Using the clarification of objective vector x arrived that module of target detection detects as the input function of SVM classifier, as f (x)=wTx+
w0>The target that target tracks for needs when 0.
The improved track algorithm is improved KCF track algorithms.KCF algorithms utilize circular matrix during tracking
And the property of SVM classifier, when being trained to grader, by the use of target basic sample as positive sample, to basic
Sample after sample loops displacement is negative sample.So only basic pattern need to originally be calculated, speed ratio is very fast.
C (x) is the matrix of a n × n, then it can be obtained by the cyclic shift of the vector x of a 1 × n, then had:
All circular matrixes can diagonalization, can be by the discrete Fourier transform of vector x:
F represents transformation constant matrix,Represent discrete Fourier transform:
The purpose of SVM classifier is to minimize:
Its solution is:
W=(XXT+λI)-1XTy
X is gained matrix after circular matrix progress discrete Fourier transform;Y represents sample label, and positive sample 1 bears sample
This is -1;The parameter of λ w over-fittings in order to control;I is unit matrix.
Bring the discrete Fourier transform of Cyclic Moment into again, the solution w of SVM classifier can also be written as:
The parameter of wherein λ over-fittings in order to control, can use 0.0001-10000, and the present embodiment takes 1;For matrix x, y pairs
Answer the discrete Fourier transform of position product.
SVM may be employed kernel function and the vector x of input be mapped in feature space φ (x).So solution of SVM classifier
W can be written as:Namely:
WhereinIt is expressed as nuclear matrix KijThe discrete Fourier transform of the first row;I represents sample number, αiIt represents to solve
W is converted to the coefficient of the linear combination of input.
Target tracking module extracts mesh using KCF algorithms using the target location that target identification module inputs as parameter
Input parameter of the target feature as SVM classifier, input parameter of the extraction clarification of objective as SVM classifier.When SVM points
Class device trains the picture that grader reads the present frame that video acquisition unit is passed to afterwards.The feature of present frame is extracted, by spy
Sign vector input grader, then:Sample labelWherein,ziIt is obtained in a new frame
Training sample, x are the object module that frame study obtains, and it is the position that maximum position is exactly target in a new frame to make y.κ tables
Show the kernel function (such as Radial basis kernel function, linear and Polynomial kernel function) with chief of a tribe's consistency.
Improved KCF algorithms use in double (more) cameras all target signatures as positive sample in target tracking module,
Sample after all positive sample cyclic shifts is negative sample, trains same SVM classifier jointly.So so that whole system
Structure is simpler, and the tracking effect in multi-cam is more robust.When track algorithm determines target in the position of a new frame
Afterwards, target tracking module can record the position data of target in each frame and be transmitted to trajectory calculation, logging modle.
The target trajectory calculates, logging modle receives the data that target tracking module is passed to, and is transmitted using tracking module
The target position information come in and the angle difference of each camera shooting calculate the three-dimensional coordinate of target in present frame.
P(Xw,YW,Zw) subpoint o'clock on the image coordinate system (pixel unit) of two cameras is P1(u1,v1)、P2
(u1,v1), it is assumed that camera C1And C2It is calibrated, M1、M2The respectively projection matrix of left and right camera, wherein M1It is as follows:
Matrix form is then had according to the conversion of three-dimensional coordinate to imaging pixel coordinate:
Wherein matrix M1For the datum after camera calibration, so we shoot pair come in image at simultaneous different cameral
It should point P1(u1,v1)、P2(u2,v2).Calculate the point P (X in three dimensionsw,YW,Zw) with the point P that is projected in two pictures1(u1,
v1)、P2(u2,v2) relation:
It is reduced to:
MsX=N
Because two pictures correspond at 2 points and same point P, M are met in space in the extended line with its camera focussIt is
Non-singular matrix, unlocking equation of n th order n can be in the hope of the world coordinates of P.
After the calculating of target trajectory computing module finishes, writing function exists the three-dimensional coordinate calculated and target
Position in image is recorded in host computer hard disk.When target trajectory calculates, after the completion of logging modle operation, startup shows mould
Block.
Display module is set including target trajectory fitting display area, displacement of targets track numerical value display area, systematic parameter
Put area.After display module starts, target following is read first, logging modle reads target trajectory calculating, logging modle record
Three-dimensional coordinate data.Then target trajectory fitting display area, displacement of targets track numerical value display area directly display out mesh
Mark the discrete points data of running orbit.Target trajectory fitting function in display module calling module afterwards, according to the mesh of reading
The discrete points data of running orbit is marked, smooth target is fitted using this non-linear fitting method of fuzzy recognition
Running orbit.Fuzzy recognition uses T-S models, and membership function is 8.
Display module system parameter setting area:The interactive interface of system, for setting the information such as measurement method, parameter.
Data management module, wherein metrical information:Including tester and testing time;Measured target information and measured object
Number;Operating position realizes the path of selection data storage file and the name of file;Tables of data measurement data is shown in specification number
According in report;The preservation of measurement data is realized in storage;Recall the inquiry for realizing measurement data;Beating for data sheet is realized in printing
Print.
As shown in Figs. 1-3, intelligent Target pursuit path record system, hardware components include host computer, double (more) visually frequency
Collecting unit and display module, the host computer receive the digital signal of video acquisition unit, and the host computer will be to receiving
Digital signal carry out intelligent recognition, tracking, trajectory calculation and record, finally export to display module and show;The video is adopted
Collection cell signal output terminal is connected with host computer;The display module receives the coordinate of host computer output object run track,
And it is output to display portion.
Host computer uses portable PC machine, dual core processor, dominant frequency 2.8GHz, memory 8GB, independent display card, video memory 512MB
Ordinary PC.The hardware deployment of whole system is facilitated, system flexibility is improved, is recorded for that can only track with target trajectory
System stable operation provides necessary condition.Storage control program in the host computer.
Claims (5)
1. intelligent Target pursuit path recording method, it is characterised in that gather target image using multiple cameras, pass through more mesh
Ranging obtains the three-dimensional track of target movement, comprises the following steps:
When each camera detect acquisition image in target for tracking target when, using KCF track algorithms to tracking target into
Line trace;
The position of target obtains the three-dimensional coordinate of target in the image gathered according to each camera;
And it records three-dimensional coordinate and is shown.
2. intelligent Target pursuit path recording method according to claim 1, it is characterised in that the KCF track algorithms bag
Include following steps:
1) the tracking target in all camera current acquired images is used as positive sample, to being obtained after positive sample cyclic shift
To negative sample;
2) using positive sample, negative sample update KCF track algorithms obtain SVM classifier in KCF track algorithms parameter w and often
Measure w0;
3) each camera inputs next frame acquisition image respectively, and the feature x of next frame acquisition image is substituted into function f respectively
(x)=wTx+w0;X be image feature, w0For the constant for plane of classifying;
Judge whether f (x) maximums of each camera are more than 0;
If it is when under preceding camera when the maximum of f (x) in preceding camera is more than the corresponding feature x position of 0, f (x) maximums
Target location in one frame acquisition image;Otherwise, then target is lost, and stops tracking;
When all cameras judge to finish, return to step 1), until target disappears or stops manually.
3. intelligent Target pursuit path recording method according to claim 2, it is characterised in that described to be cycled to positive sample
Obtaining negative sample after displacement is specially:
Positive sample with the circular matrix being made of unit matrix is multiplied, forms negative sample;
C (x) is the circular matrix of a n × n, is obtained by the cyclic shift of the vector x of a 1 × n:
4. intelligent Target pursuit path recording method according to claim 2, it is characterised in that the w passes through following steps
It obtains:
X is brought into following formulas to obtain
W=(XXT+λI)-1XTy
Wherein, X is gained matrix after circular matrix progress discrete Fourier transform;Y represents sample label, and positive sample 1 bears sample
This is -1;λ is parameter;I is unit matrix.
5. intelligent Target pursuit path recording method according to claim 1, it is characterised in that the basis each images
The position of target obtains the three-dimensional coordinate of target and is specially in the image of head acquisition:
The target location in every two field picture in each camera obtains target and the distance of setting origin in every frame;According to every
Target location in the distance of frame and the every two field picture of each camera, obtains three-dimensional coordinate of the target per frame.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611025664.9A CN108090922A (en) | 2016-11-21 | 2016-11-21 | Intelligent Target pursuit path recording method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611025664.9A CN108090922A (en) | 2016-11-21 | 2016-11-21 | Intelligent Target pursuit path recording method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108090922A true CN108090922A (en) | 2018-05-29 |
Family
ID=62169541
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611025664.9A Withdrawn CN108090922A (en) | 2016-11-21 | 2016-11-21 | Intelligent Target pursuit path recording method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108090922A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108876821A (en) * | 2018-07-05 | 2018-11-23 | 北京云视万维科技有限公司 | Across camera lens multi-object tracking method and system |
CN109407117A (en) * | 2018-09-07 | 2019-03-01 | 安徽大禹安全技术有限公司 | Earthquake emergency communication management system based on big-dipper satellite |
CN109558877A (en) * | 2018-10-19 | 2019-04-02 | 复旦大学 | Naval target track algorithm based on KCF |
CN110706291A (en) * | 2019-09-26 | 2020-01-17 | 哈尔滨工程大学 | Visual measurement method suitable for three-dimensional trajectory of moving object in pool experiment |
CN110827327A (en) * | 2018-08-13 | 2020-02-21 | 中国科学院长春光学精密机械与物理研究所 | Long-term target tracking method based on fusion |
CN111696138A (en) * | 2020-06-17 | 2020-09-22 | 北京大学深圳研究生院 | System for automatically collecting, tracking and analyzing biological behaviors |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015142923A1 (en) * | 2014-03-17 | 2015-09-24 | Carnegie Mellon University | Methods and systems for disease classification |
CN105550670A (en) * | 2016-01-27 | 2016-05-04 | 兰州理工大学 | Target object dynamic tracking and measurement positioning method |
CN106023248A (en) * | 2016-05-13 | 2016-10-12 | 上海宝宏软件有限公司 | Real-time video tracking method |
CN106127137A (en) * | 2016-06-21 | 2016-11-16 | 长安大学 | A kind of target detection recognizer based on 3D trajectory analysis |
-
2016
- 2016-11-21 CN CN201611025664.9A patent/CN108090922A/en not_active Withdrawn
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015142923A1 (en) * | 2014-03-17 | 2015-09-24 | Carnegie Mellon University | Methods and systems for disease classification |
CN105550670A (en) * | 2016-01-27 | 2016-05-04 | 兰州理工大学 | Target object dynamic tracking and measurement positioning method |
CN106023248A (en) * | 2016-05-13 | 2016-10-12 | 上海宝宏软件有限公司 | Real-time video tracking method |
CN106127137A (en) * | 2016-06-21 | 2016-11-16 | 长安大学 | A kind of target detection recognizer based on 3D trajectory analysis |
Non-Patent Citations (2)
Title |
---|
JO~AO F. HENRIQUES等: "High-Speed Tracking with Kernelized Correlation Filters", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
杨德东等: "采用核相关滤波器的长期目标跟踪", 《光学 精密工程》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108876821A (en) * | 2018-07-05 | 2018-11-23 | 北京云视万维科技有限公司 | Across camera lens multi-object tracking method and system |
CN108876821B (en) * | 2018-07-05 | 2019-06-07 | 北京云视万维科技有限公司 | Across camera lens multi-object tracking method and system |
CN110827327A (en) * | 2018-08-13 | 2020-02-21 | 中国科学院长春光学精密机械与物理研究所 | Long-term target tracking method based on fusion |
CN110827327B (en) * | 2018-08-13 | 2023-04-18 | 中国科学院长春光学精密机械与物理研究所 | Fusion-based long-term target tracking method |
CN109407117A (en) * | 2018-09-07 | 2019-03-01 | 安徽大禹安全技术有限公司 | Earthquake emergency communication management system based on big-dipper satellite |
CN109558877A (en) * | 2018-10-19 | 2019-04-02 | 复旦大学 | Naval target track algorithm based on KCF |
CN110706291A (en) * | 2019-09-26 | 2020-01-17 | 哈尔滨工程大学 | Visual measurement method suitable for three-dimensional trajectory of moving object in pool experiment |
CN111696138A (en) * | 2020-06-17 | 2020-09-22 | 北京大学深圳研究生院 | System for automatically collecting, tracking and analyzing biological behaviors |
CN111696138B (en) * | 2020-06-17 | 2023-06-30 | 北京大学深圳研究生院 | System for automatically collecting, tracking and analyzing biological behaviors |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108090922A (en) | Intelligent Target pursuit path recording method | |
CN107229930B (en) | Intelligent identification method for numerical value of pointer instrument | |
Barekatain et al. | Okutama-action: An aerial view video dataset for concurrent human action detection | |
CN103324937B (en) | The method and apparatus of label target | |
CN102521560B (en) | Instrument pointer image identification method of high-robustness rod | |
US20110025834A1 (en) | Method and apparatus of identifying human body posture | |
CN110443969A (en) | A kind of fire point detecting method, device, electronic equipment and storage medium | |
CN109190508A (en) | A kind of multi-cam data fusion method based on space coordinates | |
CN107452015A (en) | A kind of Target Tracking System with re-detection mechanism | |
CN110400315A (en) | A kind of defect inspection method, apparatus and system | |
CN109145803A (en) | Gesture identification method and device, electronic equipment, computer readable storage medium | |
CN113239797B (en) | Human body action recognition method, device and system | |
CN109886356A (en) | A kind of target tracking method based on three branch's neural networks | |
CN109919975A (en) | A kind of wide area monitoring moving target correlating method based on coordinate calibration | |
CN101826155B (en) | Method for identifying act of shooting based on Haar characteristic and dynamic time sequence matching | |
CN107038400A (en) | Face identification device and method and utilize its target person tracks of device and method | |
CN112085534B (en) | Attention analysis method, system and storage medium | |
CN105488541A (en) | Natural feature point identification method based on machine learning in augmented reality system | |
CN111399634B (en) | Method and device for recognizing gesture-guided object | |
CN108564043B (en) | Human body behavior recognition method based on space-time distribution diagram | |
CN108447092A (en) | The method and device of vision positioning marker | |
CN109492513B (en) | Face space duplication eliminating method for light field monitoring | |
CN110807375A (en) | Human head detection method, device and equipment based on depth image and storage medium | |
CN105203045B (en) | A kind of shape of product integrity detection system and inspection method based on asynchronous time domain visual sensor | |
Qin et al. | Gesture recognition from depth images using motion and shape features |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20180529 |
|
WW01 | Invention patent application withdrawn after publication |