CN106780542A - A kind of machine fish tracking of the Camshift based on embedded Kalman filter - Google Patents
A kind of machine fish tracking of the Camshift based on embedded Kalman filter Download PDFInfo
- Publication number
- CN106780542A CN106780542A CN201611246425.6A CN201611246425A CN106780542A CN 106780542 A CN106780542 A CN 106780542A CN 201611246425 A CN201611246425 A CN 201611246425A CN 106780542 A CN106780542 A CN 106780542A
- Authority
- CN
- China
- Prior art keywords
- target
- kalman filter
- observation
- camshift
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 241000251468 Actinopterygii Species 0.000 title claims abstract description 23
- 239000013598 vector Substances 0.000 claims description 52
- 239000011159 matrix material Substances 0.000 claims description 34
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 230000009897 systematic effect Effects 0.000 claims description 17
- 238000000034 method Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 9
- 239000013256 coordination polymer Substances 0.000 claims description 4
- 230000003993 interaction Effects 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000005530 etching Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000005021 gait Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000011664 nicotinic acid Substances 0.000 description 1
- 230000005311 nuclear magnetism Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
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
Abstract
The invention discloses a kind of machine fish tracking of the Camshift based on embedded Kalman filter.The complicated underwater environment of machine fish is can adapt to using the present invention, the accuracy of the quick motion tracking of machine fish is improved, real-time is good.The present invention is predicted first with Kalman filter to the position that moving target is likely to occur in next two field picture, is then scanned in the range of relative decrease using Camshift, can effectively strengthen the accuracy to fast-moving target tracking;Then, the observation for being obtained using Camshift is modified to the observation that Kalman filter is obtained, and further improves the accuracy of target following.Contrast prior art, is capable of the real-time and accuracy of Comprehensive consideration target tracking algorism, to reach the purpose that machine fish is accurately tracked in real time to target.
Description
Technical field
The present invention relates to field of intelligent control technology, and in particular to a kind of Camshift based on embedded Kalman filter
Machine fish tracking.
Background technology
Deepened continuously with to marine resources development, bionic machine fish cooperation control system causes concern, and as imitative
The important component vision subsystem of raw machine fish cooperation control system, is the unique information source of decision-making subsystem, vision
Track algorithm determines quick and precisely property and the real-time of target following.
Machine vision comes across in the collection of thesis that 1975 are edited by Winston earliest.The Marr of Britain is taught in 1973
Year at the Massachusetts Institute of Technology (MIT), a new theories of vision research group is created, in 1977, it is proposed that a kind of new
Theory on computer vision-Marr theories of vision, the theory turns into of computer vision research field in the eighties in 20th century
Highly important theoretical frame.It is used for advanced video safety monitoring technology, such as campus monitoring, traffic monitoring, road pedestrian successively
Statistical system, Visual Tracking is in missile imaging guidance, the automatically analyzing of ultrasonic wave and nuclear-magnetism sequence image, man-machine friendship in addition
Mutually, the aspect such as virtual reality, robot visual guidance has extensive important application.Every kind of NI Vision Builder for Automated Inspection be integrated with a lot, than
Such as the real-time target Detection And Tracking under static background and movement background, the Classification and Identification of target, target pose estimation, video camera
Autonomous control, video image processing system, human body gait analysis etc..It is domestic national with Institute of Automation, CAS pattern-recognition
Key lab is that many colleges and universities and research institution of representative also achieve good scientific achievement in field of machine vision.But nothing
By being domestic or external, the early stage treatment to obtained image information is all essentially consisted of in the research emphasis of field of machine vision
(denoising, enhancing, target detection etc.) and multiple video frequency motion targets are carried out in complex environment stabilization, quickly and correctly with
Track.Its difficult point is that present visual task is to recover 3D scenes by 2D images mostly, and due to there is projection in imaging process, hiding
Gear, mixing, the distortion of various scene factors etc., the objective description for wanting to obtain 3D models by the expression of highly structural is very
Difficult, these are to need asking for continuous hoisting machine vision algorithm improvement effect in current machine vision research field all the time
Topic.
In field of machine vision, the groundwork of track algorithm is that expression target area is found in continuous video sequence
Or the continuous correspondence of the picture structure of target signature.Camshift algorithms (Continuously Apative Mean-Shift)
It is a kind of motion tracking algorithms, the purpose of tracking is mainly reached by the colouring information of moving object in video image, can
According to the size for tracking change in shape self-adaptative adjustment search window of the target in motion process, the color using target is made
Be tracking characteristics, rotated in target that also there is certain robustness when being at least partially obscured, it is simple easily realize, amount of calculation is small,
Real-time is good, is an algorithm for practicality for calculating locally optimal solution.But in the unexpected quick motion conditions of complex background and target
Under be likely to occur tracking target occur error it is larger in addition tracking failure problem.
The content of the invention
In view of this, the invention provides a kind of machine fish track side of the Camshift based on embedded Kalman filter
Method, can adapt to the complicated underwater environment of machine fish, improve the accuracy of the quick motion tracking of machine fish, and real-time is good.
Camshift machine fish trackings based on embedded Kalman filter of the invention, comprise the following steps:
Step 1, before target following is carried out, by man-machine interaction mode, with mouse chosen on tracking picture manually with
Track region;
Step 2, builds Kalman filter and initializes;
Wherein, dbjective state vector x=[μ x, μ y, vx, vy]T, wherein, μ x, μ y represent the barycenter of tracked target respectively
X in the rectangular coordinate system Oxy of video image, the coordinate components in y-axis, vx, vy represent speed of the target in x, y-axis respectively
Degree component, i.e., every two frames target movement position represents transposition in x directions and the difference in y directions, subscript T;
Target observation vector y=[μ x, μ y]T;
Systematic state transfer matrixΔ t is the time difference of adjacent two frame;Systematic observation matrix
The process noise covariance matrix Q of system and the covariance matrix R of observation noise are respectively:
Wherein, e is natural logrithm;
Initial time Kalman filter Square Error matrix P0For:
Step 3, the dbjective state vector x at the Kalman filter prediction current k moment built using step 2kAnd target
Observation vector yk;
Step 4, with the dbjective state vector x of step 3 Kalman filter predictionkIn position in region of search
The heart, the target position information at current k moment is obtained using Camshift algorithms, as Camshift observation vectors yk_c;
Step 5, according to formula Kk=P'kCT·(CP'kCT+R)-1Calculate the Kalman filter gain at current k moment
COEFFICIENT Kk, wherein, P'k=APk-1AT+ Q, Pk-1It is k-1 moment Kalman filter Square Error matrix;By Kk, step 3 Kalman
The target observation vector y of filter predictionkAnd the Camshift observation vectors y that step 4 is obtainedk_c, substitute into state revision equation
x′k=xk+Kk(yk_c-yk), obtain revised state vector x 'k;And according to Pk=(1-KkC)P'kUpdate Kalman filtering equal
Square error matrix Pk;
Step 6, the revised state vector x ' that step 5 is obtainedkAs the target-like state value of present frame, read next
Two field picture, performs step 3~step 6.
Further, in the step 3, by state-transition matrix A and last moment dbjective state vector xk-1Substitution system
State equation xk=Axk-1+vk-1In, predict current k moment dbjective state vector xk, wherein, vk-1It is systematic procedure noise vector;
By predicted state vector xkSubstitute into systematic observation equation yk=Cxk+wkIn, predict current k moment target observation vector yk, wherein,
wkIt is observation noise vector.
Further, in the step 4, with the dbjective state vector x of step 3 Kalman filter predictionkIn position
As the center of the region of search at current k moment, according to the H histogram of component of previous frame image target area, in current candidate
In target area, according to Meanshift criterions, best match position is found, target centroid point is obtained, as Camshift
Observation vector yk_c。
Beneficial effect:
The present invention is carried out first with Kalman filter to the position that moving target is likely to occur in next two field picture
Prediction, is then scanned for using Camshift in the range of relative decrease, can effectively be strengthened to fast-moving target
The accuracy of tracking;Then, the observation for being obtained using Camshift is modified to the observation that Kalman filter is obtained,
Further improve the accuracy of target following.
Contrast prior art, is capable of the real-time and accuracy of Comprehensive consideration target tracking algorism, to reach machine
The purpose that fish accurately tracks in real time to target.
Brief description of the drawings
Fig. 1 is flow chart of the present invention.
Specific embodiment
Develop simultaneously embodiment below in conjunction with the accompanying drawings, and the present invention will be described in detail.
The invention provides a kind of Camshift machine fish trackings based on embedded Kalman filter, using being based on
The H histogram of component in hsv color space carries out the lookup of Meanshift algorithms to two field picture currently as matching standard performing
Before the step of target location, Kalman filter is added, completed to this frame using the target optimal location of former frame
The substantially prediction of target location in image, then makes MeanShift algorithms that searching mesh is matched in the neighborhood of this predicted position
Punctuate, the prediction using this impact point as observation to Kalman filter is modified and obtains target location optimal estimation,
The target location optimal estimation that will be obtained afterwards continues iteration carries out position prediction next time.Kalman filter only needs to know
The state vector of previous moment and the observation at current time, it is possible to deduce the state estimation of subsequent time, it is used
Recursive filtering computational methods, it is simple to initialization requirements and amount of calculation is small, so that with Camshift algorithm keeps track machines
It is more directional during device fish, so as to effectively improve the tracking accuracy to motor-driven machine fish while real-time feature is ensured.
Flow of the present invention is as shown in figure 1, mainly include 2 parts:Part i:Kalman filter is built, using former frame
Target position information and current observation in image, carry out target prodiction, and obtain the observation of Kalman target locations
Value;Part ii:Target following positioning is carried out in the target area of Kalman prediction using Camshift algorithms, is obtained
Camshift target locations observation;Ii I parts, see according to Kalman target locations observation and Camshift target locations
The difference of measured value, is modified to Kalman target prodiction values, obtains final target location.
Specifically include following steps:
Step 1, before target following is carried out, by man-machine interaction mode, with mouse chosen on tracking picture manually with
Track region;
Step 2, builds Kalman filter and initializes;
The present invention proposes the Kalman filter model of four dimensional vectors, in the rectangular coordinate system Oxy of gathered video image,
Moving situation to tracked target is predicted.
Wherein, dbjective state vector x=[μ x, μ y, vx, vy]T, wherein, μ x, μ y represent the barycenter of tracked target respectively
Coordinate components in x, y-axis, vx, vy represent velocity component of the target in x, y-axis respectively, i.e., every two frames target movement position
In x directions and the difference in y directions, subscript T represents transposition.
Target observation vector y=[μ x, μ y]T, in machine fish target observation vector, μ x, μ y are represented in present frame respectively
Coordinate components of the target centroid for observing on x, y direction.
In object tracking process, because the time interval of two continuous frames image is shorter, generally tracked target
Motion state change it is relatively small, it is possible to think tracked target adjacent two frame time interval in make uniform motion.
Correlation computations amount formula in the mathematical notation and Kalman filtering in dynamic system states space is as follows:
System state equation:xk=Axk-1+vk-1
Systematic observation equation:yk=Cxk+wk
Predicting covariance matrix:P'k=APk-1AT+Qk-1
Kalman filter gain matrix:Kk=P'kCT·(CP'kCT+Rk)-1
Kalman filter estimate:xk=Axk-1+Kk(yk-CAxk-1)
Kalman filter Square Error matrix:Pk=(1-KkC)P'k
Wherein, xk-1、xkThe dbjective state at respectively k-1, k moment is vectorial, ykIt is the target observation vector at k moment.A is to be
System state-transition matrix;vk-1It is k-1 etching process noise vectors;C is systematic observation matrix;wkIt is observation noise vector;P′kFor
K moment predicting covariance matrixes;Qk-1It is k-1 moment systematic procedure noise covariance matrixs;KkIt is Kalman filter gain
Matrix;RkIt is the covariance matrix of k moment systematic observation noises;xk-1、xkRespectively k-1, k moment Kalman filter estimate;
PkIt is k moment Kalman filter Square Error matrix.
Here the process noise vector v of systemk-1With observation noise vector wkIt is mutually independent, is all that average is zero
Gaussian sequence.It is assumed that process noise vector vk-1Obey distribution vk-1~N (0, Qk-1), wherein Qk-1It is process noise association side
Difference matrix, Qk-1=Qk=Q;Observation noise vector wkObey distribution wk~N (0, Rk), wherein RkIt is the covariance square of observation noise
Battle array, Rk-1=Rk=R.
In machine fish dbjective state vector, according to the characteristics of multiple robot fish system experimental system, corresponding ginseng is extrapolated
Number is as follows, and wherein Δ t is the time difference of adjacent two frame.
Systematic state transfer matrix A:
Systematic observation Matrix C:
By many experiments, determine that the concrete numerical value of other relevant parameters is as follows:
Systematic procedure noise covariance matrix Q:
Wherein, e is natural logrithm;
The covariance matrix R of systematic observation noise:
Setting initial time system Kalman filter Square Error matrix P0:
Step 3, according to the target-like state value of previous moment, is predicted in advance using Kalman filter, obtains current time
Dbjective state predicted value.
Specifically, by state-transition matrix A and last moment dbjective state vector xk-1Substitute into system state equation xk=
Axk-1+vk-1In, obtain the predicted state vector x of current kinetic targetk;Then by xkSubstitute into systematic observation equation yk=Cxk+wk
In, obtain the prediction observation vector y of current kinetic targetk;The related parameter values A, Q that finally previous frame is tried to achievek-1,Pk-1Together
Substitute into predicting covariance matrix equation P'k=APk-1AT+Qk-1, to predicting covariance P'kIt is predicted.
Step 4, the target predicted position x tried to achieve with step 3kIt is the center of region of search, is obtained using Camshift algorithms
The target status information at current time is obtained, as observation position yk_c(observation).
Specifically, when carrying out object matching using Camshift algorithms, the target predicted position x tried to achieve with step 3kAs
The center in current search region, according to the H histogram of component of previous frame image target area, in current candidate target area,
According to Meanshift criterions, best match position is found, obtain target centroid point yk_c=[μ x 'k,μy′k]T。
Step 5, the predicted position y obtained according to step 3 Kalman filteringkObtained using Camshift algorithms with step 4
Observation position yk_cTo correct the Kalman filtering tentative prediction of step 3, so as to obtain the target optimal estimation at current time
Position.
Specifically, first, according to formula Kk=P'kCT·(CP'kCT+Rk)-1Calculate Kalman filter gain coefficient
Kk, then by Kalman filter gain COEFFICIENT KkWith Kalman filter observation ykAnd the sight that Camshift algorithms are obtained
Location puts yk_c, substitute into state revision EQUATION x 'k=xk+Kk(yk_c-yk), obtain by the revised shape of currently practical observation information
State vector x 'k, while correcting predicting covariance matrix P'k=APk-1AT+Qk-1With Kalman filtering Square Error matrix Pk
=(1-KkC)P'k, for the target state estimator of subsequent time provides effective information.
Step 6, Kalman's correction position x ' that step 5 is obtainedkIt is the target-like state value of present frame, as in image
Target centroid point, reads next two field picture, performs step 3~step 6, repeats prediction, matching, makeover process, so that it may realize embedding
Enter tracking of the Camshift algorithms of Kalman filter to moving target.So in the treatment of each frame, Kalman is all used
Filtering carries out state estimation to target, can effectively improve the tracking effect of target.
In sum, presently preferred embodiments of the present invention is these are only, is not intended to limit the scope of the present invention.
All any modification, equivalent substitution and improvements within the spirit and principles in the present invention, made etc., should be included in of the invention
Within protection domain.
Claims (3)
1. it is a kind of based on the Camshift machine fish trackings for being embedded in Kalman filter, it is characterised in that including following step
Suddenly:
Step 1, before target following is carried out, by man-machine interaction mode, tracking area is chosen with mouse on tracking picture manually
Domain;
Step 2, builds Kalman filter and initializes;
Wherein, dbjective state vector x=[μ x, μ y, vx, vy]T, wherein, the barycenter that μ x, μ y represent tracked target respectively is being regarded
X in the rectangular coordinate system Oxy of frequency image, the coordinate components in y-axis, vx, vy represent speed of the target in x, y-axis point respectively
Amount, i.e., every two frames target movement position represents transposition in x directions and the difference in y directions, subscript T;
Target observation vector y=[μ x, μ y]T;
Systematic state transfer matrixΔ t is the time difference of adjacent two frame;Systematic observation matrix
The process noise covariance matrix Q of system and the covariance matrix R of observation noise are respectively:
Wherein, e is natural logrithm;
Initial time Kalman filter Square Error matrix P0For:
Step 3, the dbjective state vector x at the Kalman filter prediction current k moment built using step 2kWith target observation to
Amount yk;
Step 4, with the dbjective state vector x of step 3 Kalman filter predictionkIn position for region of search center, utilize
Camshift algorithms obtain the target position information at current k moment, as Camshift observation vectors yk_c;
Step 5, according to formula Kk=P'kCT·(CP'kCT+R)-1Calculate the Kalman filter gain coefficient at current k moment
Kk, wherein, P'k=APk-1AT+ Q, Pk-1It is k-1 moment Kalman filter Square Error matrix;By Kk, step 3 Kalman filtering
The target observation vector y of device predictionkAnd the Camshift observation vectors y that step 4 is obtainedk_c, substitute into state revision EQUATION x 'k=
xk+Kk(yk_c-yk), obtain revised state vector x 'k;And according to Pk=(1-KkC)P'kUpdate Kalman filtering mean square error
Difference matrix Pk;
Step 6, the revised state vector x ' that step 5 is obtainedkAs the target-like state value of present frame, next frame figure is read
Picture, performs step 3~step 6.
2. the Camshift machine fish trackings of embedded Kalman filter are based on as claimed in claim 1, and its feature exists
In in the step 3, by state-transition matrix A and last moment dbjective state vector xk-1Substitute into system state equation xk=
Axk-1+vk-1In, predict current k moment dbjective state vector xk, wherein, vk-1It is systematic procedure noise vector;By predicted state
Vector xkSubstitute into systematic observation equation yk=Cxk+wkIn, predict current k moment target observation vector yk, wherein, wkFor observation is made an uproar
Sound vector.
3. the Camshift machine fish trackings of embedded Kalman filter are based on as claimed in claim 1, and its feature exists
In in the step 4, with the dbjective state vector x of step 3 Kalman filter predictionkIn position as the current k moment
The center of region of search, according to the H histogram of component of previous frame image target area, in current candidate target area, foundation
Meanshift criterions, find best match position, target centroid point are obtained, as Camshift observation vectors yk_c。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611246425.6A CN106780542A (en) | 2016-12-29 | 2016-12-29 | A kind of machine fish tracking of the Camshift based on embedded Kalman filter |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611246425.6A CN106780542A (en) | 2016-12-29 | 2016-12-29 | A kind of machine fish tracking of the Camshift based on embedded Kalman filter |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106780542A true CN106780542A (en) | 2017-05-31 |
Family
ID=58929010
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611246425.6A Pending CN106780542A (en) | 2016-12-29 | 2016-12-29 | A kind of machine fish tracking of the Camshift based on embedded Kalman filter |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106780542A (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107480602A (en) * | 2017-07-26 | 2017-12-15 | 哈尔滨工业大学深圳研究生院 | A kind of inexpensive target visual tracking and system |
CN108038868A (en) * | 2017-10-17 | 2018-05-15 | 国网河南省电力公司郑州供电公司 | Across the visual field method for tracking target of substation's complex environment based on three-dimensional digital model |
CN108492324A (en) * | 2018-01-27 | 2018-09-04 | 天津大学 | Aircraft method for tracing based on fully-connected network and Kalman filter |
CN109063600A (en) * | 2018-07-13 | 2018-12-21 | 河南大学 | Human motion method for tracing and device based on face recognition |
CN110443832A (en) * | 2019-06-21 | 2019-11-12 | 西北工业大学 | A kind of evidence filtered target tracking based on observation interval value |
CN110555377A (en) * | 2019-07-27 | 2019-12-10 | 华南理工大学 | pedestrian detection and tracking method based on fisheye camera overlook shooting |
CN110706253A (en) * | 2019-09-19 | 2020-01-17 | 中国科学院自动化研究所 | Target tracking method, system and device based on apparent feature and depth feature |
CN110876275A (en) * | 2019-04-30 | 2020-03-10 | 深圳市大疆创新科技有限公司 | Aiming control method, mobile robot and computer readable storage medium |
CN110986895A (en) * | 2019-12-06 | 2020-04-10 | 昆明理工大学 | Method and system for judging complete water entry of lead fish based on Kalman filtering algorithm |
CN111612729A (en) * | 2020-05-26 | 2020-09-01 | 杭州电子科技大学 | Target sequence tracking image recovery method based on Kalman filtering |
CN111724405A (en) * | 2020-06-01 | 2020-09-29 | 厦门大学 | Long-time multi-target prawn tracking method based on boundary constraint Kalman filtering |
CN113283380A (en) * | 2021-06-11 | 2021-08-20 | 张洁欣 | Children motion attitude automatic identification method based on 3D convolution long-term and short-term memory network |
CN113709324A (en) * | 2020-05-21 | 2021-11-26 | 武汉Tcl集团工业研究院有限公司 | Video noise reduction method, video noise reduction device and video noise reduction terminal |
CN113810854A (en) * | 2021-09-16 | 2021-12-17 | 中国联合网络通信集团有限公司 | Terminal motion track determination method and server |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101794385A (en) * | 2010-03-23 | 2010-08-04 | 上海交通大学 | Multi-angle multi-target fast human face tracking method used in video sequence |
CN102737385A (en) * | 2012-04-24 | 2012-10-17 | 中山大学 | Video target tracking method based on CAMSHIFT and Kalman filtering |
CN104318211A (en) * | 2014-10-17 | 2015-01-28 | 中国传媒大学 | Anti-shielding face tracking method |
US20150178568A1 (en) * | 2013-12-23 | 2015-06-25 | Canon Kabushiki Kaisha | Method for improving tracking using dynamic background compensation with centroid compensation |
-
2016
- 2016-12-29 CN CN201611246425.6A patent/CN106780542A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101794385A (en) * | 2010-03-23 | 2010-08-04 | 上海交通大学 | Multi-angle multi-target fast human face tracking method used in video sequence |
CN102737385A (en) * | 2012-04-24 | 2012-10-17 | 中山大学 | Video target tracking method based on CAMSHIFT and Kalman filtering |
US20150178568A1 (en) * | 2013-12-23 | 2015-06-25 | Canon Kabushiki Kaisha | Method for improving tracking using dynamic background compensation with centroid compensation |
CN104318211A (en) * | 2014-10-17 | 2015-01-28 | 中国传媒大学 | Anti-shielding face tracking method |
Non-Patent Citations (1)
Title |
---|
李明锁 等: "《结合扩展卡尔曼滤波的CamShift移动目标跟踪算法》", 《电光与控制》 * |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107480602B (en) * | 2017-07-26 | 2021-04-30 | 哈尔滨工业大学深圳研究生院 | Low-cost target visual tracking method and system |
CN107480602A (en) * | 2017-07-26 | 2017-12-15 | 哈尔滨工业大学深圳研究生院 | A kind of inexpensive target visual tracking and system |
CN108038868A (en) * | 2017-10-17 | 2018-05-15 | 国网河南省电力公司郑州供电公司 | Across the visual field method for tracking target of substation's complex environment based on three-dimensional digital model |
CN108492324A (en) * | 2018-01-27 | 2018-09-04 | 天津大学 | Aircraft method for tracing based on fully-connected network and Kalman filter |
CN108492324B (en) * | 2018-01-27 | 2021-05-11 | 天津大学 | Airplane tracking method based on full-connection network and Kalman filter |
CN109063600A (en) * | 2018-07-13 | 2018-12-21 | 河南大学 | Human motion method for tracing and device based on face recognition |
CN110876275A (en) * | 2019-04-30 | 2020-03-10 | 深圳市大疆创新科技有限公司 | Aiming control method, mobile robot and computer readable storage medium |
CN110443832A (en) * | 2019-06-21 | 2019-11-12 | 西北工业大学 | A kind of evidence filtered target tracking based on observation interval value |
CN110443832B (en) * | 2019-06-21 | 2022-03-01 | 西北工业大学 | Evidence filtering target tracking method based on observation interval value |
CN110555377A (en) * | 2019-07-27 | 2019-12-10 | 华南理工大学 | pedestrian detection and tracking method based on fisheye camera overlook shooting |
CN110555377B (en) * | 2019-07-27 | 2023-06-23 | 华南理工大学 | Pedestrian detection and tracking method based on fish eye camera overlooking shooting |
CN110706253A (en) * | 2019-09-19 | 2020-01-17 | 中国科学院自动化研究所 | Target tracking method, system and device based on apparent feature and depth feature |
CN110706253B (en) * | 2019-09-19 | 2022-03-08 | 中国科学院自动化研究所 | Target tracking method, system and device based on apparent feature and depth feature |
CN110986895A (en) * | 2019-12-06 | 2020-04-10 | 昆明理工大学 | Method and system for judging complete water entry of lead fish based on Kalman filtering algorithm |
CN113709324A (en) * | 2020-05-21 | 2021-11-26 | 武汉Tcl集团工业研究院有限公司 | Video noise reduction method, video noise reduction device and video noise reduction terminal |
CN111612729A (en) * | 2020-05-26 | 2020-09-01 | 杭州电子科技大学 | Target sequence tracking image recovery method based on Kalman filtering |
CN111612729B (en) * | 2020-05-26 | 2023-06-23 | 杭州电子科技大学 | Target sequence tracking image recovery method based on Kalman filtering |
CN111724405A (en) * | 2020-06-01 | 2020-09-29 | 厦门大学 | Long-time multi-target prawn tracking method based on boundary constraint Kalman filtering |
CN113283380A (en) * | 2021-06-11 | 2021-08-20 | 张洁欣 | Children motion attitude automatic identification method based on 3D convolution long-term and short-term memory network |
CN113810854A (en) * | 2021-09-16 | 2021-12-17 | 中国联合网络通信集团有限公司 | Terminal motion track determination method and server |
CN113810854B (en) * | 2021-09-16 | 2023-10-03 | 中国联合网络通信集团有限公司 | Method for determining motion trail of terminal and server |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106780542A (en) | A kind of machine fish tracking of the Camshift based on embedded Kalman filter | |
CN107292911B (en) | Multi-target tracking method based on multi-model fusion and data association | |
Cannons | A review of visual tracking | |
US7706571B2 (en) | Flexible layer tracking with weak online appearance model | |
CN109949346B (en) | Weld joint tracking method based on direction gradient histogram particle filtering | |
Choi et al. | Robust 3D visual tracking using particle filtering on the SE (3) group | |
CN107742306B (en) | Moving target tracking algorithm in intelligent vision | |
Madasu et al. | Estimation of vehicle speed by motion tracking on image sequences | |
CN109448023B (en) | Satellite video small target real-time tracking method | |
CN106780560A (en) | A kind of feature based merges the bionic machine fish visual tracking method of particle filter | |
Ait Abdelali et al. | An adaptive object tracking using Kalman filter and probability product kernel | |
CN109902543A (en) | Target trajectory estimation method, device and Target Tracking System | |
CN103337082A (en) | Video segmentation method based on statistical shape prior | |
Kwolek et al. | Swarm intelligence based searching schemes for articulated 3D body motion tracking | |
Chen et al. | Real-time Visual Object Tracking via CamShift-Based Robust Framework. | |
Nickel et al. | Dynamic integration of generalized cues for person tracking | |
Ghedia et al. | A novel approach for monocular 3d object tracking in cluttered environment | |
Martínez et al. | Tracking human body parts using particle filters constrained by human biomechanics | |
Zhuo et al. | 3D Visual Motion Amplitude Tracking Simulation Method for Sports | |
Pawar et al. | Effective Object Detection and Tracking for Holonomic Robot using Deep Neural Architecture | |
Liu et al. | Computer Vision-Based Tracking of Workers in Construction Sites Based on MDNet | |
Gong et al. | Real-time Visual SLAM based YOLO-Fastest for Dynamic Scenes | |
Hao et al. | Moving object tracking algorithm based on improved gaussian mixture model | |
Yan et al. | Model-based 3D human motion tracking and voxel reconstruction from sparse views | |
CN117670931A (en) | Multi-camera multi-target tracking method and device based on computer vision |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170531 |
|
RJ01 | Rejection of invention patent application after publication |