CN101226592A - Method for tracing object based on component - Google Patents
Method for tracing object based on component Download PDFInfo
- Publication number
- CN101226592A CN101226592A CNA200810033733XA CN200810033733A CN101226592A CN 101226592 A CN101226592 A CN 101226592A CN A200810033733X A CNA200810033733X A CN A200810033733XA CN 200810033733 A CN200810033733 A CN 200810033733A CN 101226592 A CN101226592 A CN 101226592A
- Authority
- CN
- China
- Prior art keywords
- parts
- tracking
- tracking unit
- point
- grey level
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
A component-based object tracking method in the field of image processing techniques comprises steps of firstly, for appeared tracking objects, positioning tracking components of the tracking objects by employing a method of speeding area corner detection, then describing the tracking components through a gray histogram, further tacking the tracking components in subsequent frames through Kalman filtering, then correcting parameter of the Kalman filtering in each frame through measuring the observed value and updating the components and lastly, identifying the tracking objects. The invention can comparatively accurately track the objects with quite speedy tracking speed, and the invention utilizes a small widow which takes corner points of the objects as centers, as tracking components, which is capable of effectively overcoming the problem of shielding and the like.
Description
Technical field
The present invention relates to a kind of method of technical field of image processing, specifically is a kind of object tracking based on parts.
Background technology
The object tracking problem is important field of research in the computer vision, and it all has in fields such as video monitoring, compression of images and three-dimensional reconstruction widely uses.The motion object the motion process in to run into the problem of blocking unavoidably, blocking is the principal element that influences tracking stability, overcome block be track algorithm difficult point it
Find by prior art documents, (in January, 2006: the Robust Fragments-based Tracking using the Integral Histogram (based on parts and histogrammic tracking) that 798-805) delivers proposes to follow the tracks of window and is divided into the experimental process window Amit Adam etc. at Computer Vision andPattern Recognition (computer vision and Model Identification), recompense for the subwindow histogram that is blocked, but its deficiency is to adopt the method for poor search to mate the real-time that two windows have reduced tracking, and the result can not embody the movement angle of tracing object.
Summary of the invention
The objective of the invention is to overcome above-mentioned the deficiencies in the prior art, a kind of tracking based on parts has been proposed, with a plurality of parts in the target as tracing object, it is adopted based on the grey level histogram of nuclear describe each parts in the tracing object, parameter by Kalman filter prediction parts, then utilize based on the grey level histogram of nuclear and revise, to finish tracking, not only overcome occlusion issue effectively, and overcome problems such as inner relative motion that exists of object and non-rigid body distortion, have good real time performance and good tracking effect.
The present invention is achieved by the following technical solutions, comprises the steps:
At first to the tracing object of appearance, the method for use FAST (acceleration region) Corner Detection positions the tracking unit of tracing object;
Describe tracking unit by grey level histogram then, in follow-up frame, tracking unit is followed the tracks of, in every frame, revise the parameter of Kalman filtering, and carry out the renewal of parts by the measurement of observed reading by Kalman filtering;
At last tracing object is identified out.
Described tracking unit to tracing object positions, be meant: the method for employing FAST Corner Detection detects the angle point in the motion object, the rectangular window that will be the center with the angle point is as tracking unit, the method of FAST Corner Detection is specially: for checking measuring point c to be checked circle on every side, seek wherein the longest circular arc, if gray-scale value t that the gray-scale value of all points is all ordered greater than c in the circular arc more than the gray-scale value, t is set as required by the user, gray-scale value t that perhaps all orders less than c more than the gray-scale value, then be judged as angle point.
Describedly tracking unit is described by grey level histogram, be meant: adopt based on the grey level histogram of nuclear and describe tracking unit, histogram is a n-dimensional vector, n is set as required by the user, at first color space is mapped to the n dimension by 256 dimensions, adopts Biweight (two weight) kernel function that each is put weighting then, make the weights of distance center pixel far away less, can reduce the influence of ground unrest like this, improve histogrammic stability.
Describedly tracking unit is followed the tracks of by Kalman filtering, be specially: Kalman filtering comprises prediction and revises two parts: predicted portions adopts the predictive equation group, utilize the state value and the predicated error of previous moment to make prediction, obtain the position of each tracking unit at current time, can there be certain error owing to predict the outcome, revise part and adopt the update equation group, utilize the observed reading correction of the current time that obtains to predict the outcome.
The parameter of Kalman filter is revised in described measurement by observed reading, be meant: the method that around the tracking unit position that the predictive equation group dopes, adopts spiral search, find a bit, with the histogram of this some window that is neighborhood and the histogrammic Euclidean distance of former parts less than preset threshold, the position of this point is just as the observed reading of present frame, the update equation group is revised the predicted value of current Kalman filter by this observed reading, obtains revised state estimation value and Noise Variance Estimation value.
The described renewal of carrying out parts, be specially: if around future position, can't find qualified point, adopt decision method with object identity, keep the parts that are positioned within the identified areas, promptly keep because object rotation or owing to block the parts disappearance that causes but still parts within tracing object, eliminate because of following the tracks of failure and be in parts outside the tracing object, and the method for the grey level histogram weighting summation of the grey level histogram of former parts and current time parts present position is upgraded grey level histogram.
Described tracing object is identified out, be meant: determine after each parts of tracing object, rectangle with the area minimum that can comprise all tracking unit central points identifies out with tracing object, make each parts unification to an object, be specially: adopt the Graham Sodd method of investing method to determine the convex hull of object angle point set, obtain after the convex hull of point set, extended line with a limit on the convex hull is one side of rectangle, find out and comprise a little the rectangle of concentrating all points, the rectangular area of area minimum is found out in rotation successively.
Compared with prior art, the present invention has following beneficial effect: the present invention is tracking target and have very fast tracking velocity more exactly, use the object angle point as the wicket at center as tracking unit, can overcome problem such as block effectively.Object factory adopts the grey level histogram based on nuclear, with the situation of adaption object size variation and rotation, and the influence of reduction ground unrest.By the parameter of Kalman filtering prediction parts, and adopt the histogram modification filter parameter, guaranteed the accuracy that parts are followed the tracks of at next frame.Determine the rectangle at object centers place at last with the Graham Sodd method of investing method, make rectangle inclination angle and size all comparatively meet the features of shape of target.And the present invention has good real time performance and stable tracking effect, and can overcome effectively block, problem such as object medium velocity disunity.
Description of drawings
Fig. 1 is a FAST angular-point detection method synoptic diagram in the embodiments of the invention;
The process flow diagram of Kalman filtering among Fig. 2 the present invention;
Fig. 3 is that the present invention determines minimum rectangle area synoptic diagram;
The initial position of parts and the location drawing in subsequent frame in the embodiments of the invention experiment of Fig. 4 position;
Fig. 5 is the figure as a result of experiment one in the embodiments of the invention;
Fig. 6 is the figure as a result of experiment two in the embodiments of the invention;
Fig. 7 blocks result of experiment figure in the embodiments of the invention;
Fig. 8 is the effect comparison diagram of the present invention and mean shift tracking.
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
What present embodiment adopted is the moving object of following the tracks of in one section video sequence, comprises following concrete steps:
1, positioning parts: it is the rectangular area at center that the tracking unit in the present embodiment is chosen in the tracing object with the angle point, Corner Detection adopts the FAST angular-point detection method, FAST is a kind of computing simple and direct angular-point detection method, can guarantee the requirement of tracker real-time.The angle point that the FAST angular-point detection method detects is for having enough points and this point to be in different zones around this point, be specially: check measuring point c to be checked circle on every side, seek wherein the longest circular arc, if gray-scale value t that the gray-scale value of all points is all ordered greater than c in the circular arc more than the gray-scale value (t=15), gray-scale value t that perhaps all orders less than c more than the gray-scale value, then be judged as angle point.
As shown in Figure 1, if the gray-scale value of relative 2 points (such as picture element 1 and 9) is all close with the gray-scale value that c is ordered on the circumference, does not so obviously need to detect 12 points and judge whether c is angle point, therefore can be optimized said method, detect pixel 1,9 earlier, detect 5 and 13 then.Add up by experiment, after optimizing, for a point to be detected in the image, whether average 3.8 points that need detect around it just can be judged is angle point.
2, use based on the grey level histogram vector of examining
Describe tracking unit, suppose { x
i *}
I=1...mAround the expression parts center angle point c be with the radius in the circle of r all pixels with respect to the normalization coordinate of c, the distribution of color of object is turned to the n dimension by 256 original dimensions, n is set at 16 in the present embodiment, 's the one dimension of pixel mapping in the n dimension color space at its place of x, then grey level histogram by function b (x) with the position
Computing formula be:
Wherein: b (x)=I (x)/n, I (x) expression position is the gray-scale value of the pixel of x; K (x) is a kernel function, in the present embodiment, adopts the Biweight kernel function, and its expression formula is
Make the weights of distance center pixel far away less, reduce the influence of ground unrest, improved histogrammic stability.
3, as shown in Figure 2, adopt Kalman filtering that tracking unit is followed the tracks of, the tracking step of Kalman filtering is divided into prediction and revises two parts:
Predicted portions is: the predictive equation group is utilized the state value of previous moment and the predicted value that predicated error obtains current time, and the predictive equation group is:
x
k'=Ax
K-1+ Bu
K-1(B=0 in the present embodiment)
P
k′＝AP
k-1A
T+Q
Wherein: x
k' be k predicted state variable constantly, x
K-1Be k-1 state variable constantly, A is a state-transition matrix, P
k' be k predicated error correlation matrix constantly, P
K-1Be the modified value of k-1 moment error correlation matrix, Q is the motion noise correlation matrix.
Retouch is: obtain after predicted value and the prediction correlated error, the update equation group is made correction by the observed reading of current time to predicted value and prediction correlated error, and the update equation group is:
K
k＝P
k′H
T(HP
k′H
T+R)
-1
x
k＝x
k′+K
k(z
k-Hx
k′)
P
k＝(I-K
kH)P
k′
Wherein: K
kBe the Kalman filtering gain matrix, H is for measuring matrix, and R is for measuring noise correlation matrix, z
kObservational variable, P
kBe the modified value of k moment error correlation matrix, x
kBe k state variable constantly.
The update equation group is by observed reading z
kRevise current predicted value, obtain revised state estimation value and Noise Variance Estimation value.
In the present embodiment, observed reading is set at the position at place, tracking unit center, and state variable is made as position, movement velocity and the acceleration at parts center, and the hypothesis parts do uniformly accelerated motion, then that parameter setting is as follows:
Z=[s
x, s
y]
T, x=[s
x, v
x, a
x, s
y, v
y, a
y]
T, (z is an observed reading, and subscript x, y represent horizontal ordinate and ordinate)
State-transition matrix
Measure matrix
Parameter P
k, motion noise correlation matrix Q and measure noise correlation matrix R initial value and need utilize priori to determine, to fall into parts window optional position all be equiprobable to hypothetic observation in the present embodiment, the initial value of setting measurement noise correlation matrix R is:
N
x, N
yBe respectively the pixel count of the parts window transverse axis and the longitudinal axis.
4. the measurement of observed reading and component update: the displacement coordinate (s that is doping
x', s
y') for taking the method for spiral search in the small neighbourhood at center, find the coordinate of observed reading apace, search first and meet ρ [p spiral
k, q
K-1The coordinate of]＜l is as z
k, ρ [p wherein
k, q
K-1] be the distance of the histogram and the object color model of current coordinate, l is a distance threshold.
If can't find qualified point around the future position, mainly due to following three kinds of reasons: (1) these parts have been followed the tracks of failure, and promptly parts have been in outside the tracing object; (2) owing to reasons such as object rotations, parts disappear, but follow the tracks of window still within tracing object; (3) owing to occurred blocking, parts are disappeared.For stability that guarantees to follow the tracks of and the parts that abundant quantity is arranged, need to eliminate the parts under first kind of situation and keep back two kinds of parts.The decision method that present embodiment is taked is positioned at the parts reservation within the identified areas, otherwise eliminates for after the object identity.
In addition, because tracing object is in the continuous variation, and there is illumination variation, noise in the image, problem such as blocks, so will its color model be upgraded, present embodiment upgrades color model with the histogram weighting of the observed reading of present frame, and more new formula is as follows: q
k=(1-α) q
K-1+ α p
k, wherein: q
k, q
K-1Be respectively the grey level histogram based on nuclear of k, k-1 frame, α is set at 0.02 for upgrading the factor in the present embodiment.
5. sign object: determine after each parts of tracing object, rectangle with the area minimum that can comprise all tracking unit central points identifies out with tracing object, make each parts unification to an object, because there is abundant angle point in target edges, the method can comparatively accurately identify size, position and the inclination angle of object.
Must at first determine the convex hull of angle point point set for the rectangle of seeking the area minimum that can comprise all tracking unit central points.Present embodiment adopts the Graham Sodd method of investing method to determine the convex hull of object angle point set, the convex hull of plane point set is defined as the minimum convex set that comprises point set, be a convex polygon on summit promptly with point set middle part branch, to any limit of this convex polygon, point concentrate all not the point on this limit all in the same side on this limit.
Described Graham Sodd method of investing method (Zhou Peide. " computational geometry-Algorithmic Design ﹠ Analysis (second edition) ". the .2005. of publishing house of Tsing-Hua University), be specially:
1. establish that the point of coordinate minimum is p in the convex set
1, p
1Connect with line segment with other each points in the convex set, and calculate these line segments and horizontal angle, press corner dimension then and arrive p
1Distance sort, obtain a sequence p
1, p
2... p
n, p
1Point is the summit on convex hull border, p
2With p
nAlso must be;
2. judge whether it is the point of convex hull, leave out p
3, p
4..., p
N-1In be not point on the convex hull, specific as follows:
[1] sets k=4;
[2] set j=2;
[3] if p
1And p
kRespectively at line segment p
K-j+1p
K-jP is then left out in both sides
K-j+1, the descendant vertex numbering subtracts 1, k=k-1, j=j-1, n=n-1; Otherwise p
K-j+1Be the convex hull summit temporarily, and record;
[4] j=j+1 carries out [3], up to j=k-2;
[5] k=k+1 carries out [2], up to k=n.
Wherein: judge at 2 at some line segments both sides or homonymy with vectorial multiplication cross, than: 2 P are arranged, Q, line segment AB calculates multiplication cross PA * PB, and QA * QB is if the jack per line explanation is at the line segment homonymy, otherwise at heteropleural.
3. order is exported the convex hull summit;
Obtaining after the convex hull of point set, is one side of rectangle with the extended line on a limit on the convex hull, finds out to comprise the rectangle of a little concentrating all points, and the rectangular area of area minimum is found out in rotation successively.
As shown in Figure 3, supposing that the convex hull summit of determining is ABCDE, then is that the rectangular area on limit is with AB:
In like manner, can calculate with BC, CD, DE, EA is the rectangular area on limit, and the rectangle of therefrom selecting the area minimum identifies tracing object.
Shown in Fig. 5,6,7, three experimental result pictures for present embodiment, as seen from the figure, the present embodiment tracking in video sequence, exist a large amount of owing under the situation of the noise that illumination variation and camera shake bring, still can be described out position, the size of object and the angle of rotating accurately.In the process that target draws near and turns, stable tracking results is arranged still.
As shown in Figure 4, figure (a) is the position in subsequent frame for the initial position of the parts of experiment 1 and experiment 2, figure (b);
As shown in Figure 5, for testing 1 figure as a result, (a) (b) (c) (d) is respectively the image of 185 frames, 259 frames, 574 frames, 704 frames, there is a people from car, to walk out in the 574th frame among the figure (c), among the figure (d) in the 704th frame this person prove that the present embodiment method can overcome to block and recovery from block away from the vehicle of following the tracks of.
As shown in Figure 6, for testing 2 figure as a result, figure (a) (b) (c) (d) is respectively the image of 38 frames, 91 frames, 118 frames, 171 frames, the background trees swing with the wind among the figure, and each position speed of following the tracks of personage's health is also inconsistent, finds out that thus the method can overcome problems such as complicated variation of background and tracing object internal speed disunity.
As shown in Figure 7, for blocking experiment, artificially added pixel value in the experiment and be 250 occlusion area, figure (a) (b) is respectively the image of 230 frames, 260 frames, and experimental result present embodiment method as can be seen has the stable anti-ability of blocking.
As shown in table 1, for the velocity ratio of present embodiment method and particle filter tracking method, the speed of present embodiment method improves a lot, for further target being discerned and analyzed the assurance that real-time is provided.
Table 1 be present embodiment method and particle filter tracking method velocity ratio;
| Tracking | The present embodiment method | Particle filter | ||
| Performance parameter | Frame number | Component count/frame | Millisecond/frame | Millisecond/frame |
| Experiment 1 | 1302 | 38 | 3.04 | 62.3 |
| Experiment 2 | 150 | 51 | 2.92 | 59.7 |
As shown in Figure 8, the figure (a) and (b) are the design sketch of embodiment method, figure (c), (d) are the design sketch of Mean-Shift (mean shift) tracking, the present embodiment method is compared with the mean shift method, this method is localizing objects more accurately, change the size and the anglec of rotation of following the tracks of window more flexibly, and have higher stability.
Claims (8)
1. object tracking based on parts, it is characterized in that, comprise the steps: at first tracing object to occurring, use the acceleration region angular-point detection method that the tracking unit of tracing object is positioned, describe tracking unit by grey level histogram then, in follow-up frame, tracking unit is followed the tracks of, in every frame, revise the parameter of Kalman filtering by the measurement of observed reading by Kalman filtering, and carry out the renewal of parts, at last tracing object is identified out.
2. the object tracking based on parts according to claim 1, it is characterized in that, described tracking unit to tracing object positions, be meant: adopt the angle point in the acceleration region angular-point detection method detection motion object, the rectangular window that will be the center with the angle point is as tracking unit, the method of acceleration region Corner Detection is specially: for checking measuring point c to be checked circle on every side, seek wherein the longest circular arc, if gray-scale value t that the gray-scale value of all points is all ordered greater than c in the circular arc more than the gray-scale value, t is set as required by the user, gray-scale value t that perhaps all orders less than c more than the gray-scale value, then be judged as angle point.
3. the object tracking based on parts according to claim 1, it is characterized in that, describedly tracking unit is described by grey level histogram, be meant: adopt based on the grey level histogram of nuclear and describe tracking unit, histogram is a n-dimensional vector, and n is set as required by the user, at first color space is mapped to the n dimension by 256 dimensions, adopt two weight kernel functions that each is put weighting then, make the weights of distance center pixel far away less.
4. the object tracking based on parts according to claim 1, it is characterized in that, describedly tracking unit is followed the tracks of by Kalman filtering, be specially: Kalman filtering comprises prediction and revises two parts: predicted portions adopts the predictive equation group, utilize the state value and the predicated error of previous moment to make prediction, obtain the position of each tracking unit at current time, can there be certain error owing to predict the outcome, revise part and adopt the update equation group, utilize the observed reading correction of the current time that obtains to predict the outcome.
5. the object tracking based on parts according to claim 1, it is characterized in that, the parameter of Kalman filter is revised in described measurement by observed reading, be meant: the method that around the tracking unit position that the predictive equation group dopes, adopts spiral search, find a bit, with the histogram of this some window that is neighborhood and the histogrammic Euclidean distance of former parts less than preset threshold, the position of this point is just as the observed reading of present frame, the update equation group is revised the predicted value of current Kalman filter by this observed reading, obtains revised state estimation value and Noise Variance Estimation value.
6. the object tracking based on parts according to claim 1, it is characterized in that, the renewal of described parts, be specially: if around future position, can't find qualified point, adopt decision method with object identity, keep the parts that are positioned within the identified areas, promptly keep because object rotation or owing to block the parts disappearance that causes but still parts within tracing object, eliminate because of following the tracks of failure and be in parts outside the tracing object, and the method for the grey level histogram weighting summation of the grey level histogram of former parts and current time parts present position is upgraded grey level histogram.
7. according to claim 1 or 6 described object trackings, it is characterized in that based on parts, the renewal of described parts, more new formula is as follows for it: q
k=(1-α) q
K-1+ α p
k, wherein: q
k, q
K-1Be respectively the grey level histogram based on nuclear of k, k-1 frame, α is for upgrading the factor.
8. the object tracking based on parts according to claim 1, it is characterized in that, described tracing object is identified out, be meant: determine after each parts of tracing object, rectangle with the area minimum that can comprise all tracking unit central points identifies out with tracing object, make each parts unification to an object, be specially: adopt the Graham Sodd method of investing method to determine the convex hull of object angle point set, obtain after the convex hull of point set, extended line with a limit on the convex hull is one side of rectangle, find out and comprise a little the rectangle of concentrating all points, the rectangular area of area minimum is found out in rotation successively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200810033733XA CN101226592A (en) | 2008-02-21 | 2008-02-21 | Method for tracing object based on component |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200810033733XA CN101226592A (en) | 2008-02-21 | 2008-02-21 | Method for tracing object based on component |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101226592A true CN101226592A (en) | 2008-07-23 |
Family
ID=39858577
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA200810033733XA Pending CN101226592A (en) | 2008-02-21 | 2008-02-21 | Method for tracing object based on component |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101226592A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101727570B (en) * | 2008-10-23 | 2012-05-23 | 华为技术有限公司 | Tracking method, track detection processing unit and monitor system |
| CN102930557A (en) * | 2012-10-16 | 2013-02-13 | 苏州大学 | Particle filter tracking method for adaptive adjustment of tracking window size |
| CN102930245A (en) * | 2012-09-24 | 2013-02-13 | 深圳市捷顺科技实业股份有限公司 | Method and system for tracking vehicles |
| CN103106659A (en) * | 2013-01-28 | 2013-05-15 | 中国科学院上海微系统与信息技术研究所 | Open area target detection and tracking method based on binocular vision sparse point matching |
| CN103279952A (en) * | 2013-05-17 | 2013-09-04 | 华为技术有限公司 | Target tracking method and device |
| CN105898196A (en) * | 2014-11-24 | 2016-08-24 | 北京高尔智达科技有限公司 | Multi-spectral photoelectric automatic recognition and tracking system |
| CN106683120A (en) * | 2016-12-28 | 2017-05-17 | 杭州趣维科技有限公司 | Image processing method being able to track and cover dynamic sticker |
| CN106952295A (en) * | 2017-03-17 | 2017-07-14 | 公安部第三研究所 | A kind of implementation method of the rotor wing unmanned aerial vehicle pursuit movement target of view-based access control model |
| CN107851318A (en) * | 2015-08-18 | 2018-03-27 | 高通股份有限公司 | System and method for Object tracking |
| CN110633731A (en) * | 2019-08-13 | 2019-12-31 | 杭州电子科技大学 | Single-stage anchor-frame-free target detection method based on staggered sensing convolution |
| CN115037869A (en) * | 2021-03-05 | 2022-09-09 | Oppo广东移动通信有限公司 | Automatic focusing method and device, electronic equipment and computer readable storage medium |
-
2008
- 2008-02-21 CN CNA200810033733XA patent/CN101226592A/en active Pending
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101727570B (en) * | 2008-10-23 | 2012-05-23 | 华为技术有限公司 | Tracking method, track detection processing unit and monitor system |
| CN102930245A (en) * | 2012-09-24 | 2013-02-13 | 深圳市捷顺科技实业股份有限公司 | Method and system for tracking vehicles |
| CN102930245B (en) * | 2012-09-24 | 2015-03-18 | 深圳市捷顺科技实业股份有限公司 | Method and system for tracking vehicles |
| CN102930557A (en) * | 2012-10-16 | 2013-02-13 | 苏州大学 | Particle filter tracking method for adaptive adjustment of tracking window size |
| CN103106659A (en) * | 2013-01-28 | 2013-05-15 | 中国科学院上海微系统与信息技术研究所 | Open area target detection and tracking method based on binocular vision sparse point matching |
| CN103279952A (en) * | 2013-05-17 | 2013-09-04 | 华为技术有限公司 | Target tracking method and device |
| CN103279952B (en) * | 2013-05-17 | 2017-10-17 | 华为技术有限公司 | A kind of method for tracking target and device |
| CN105898196A (en) * | 2014-11-24 | 2016-08-24 | 北京高尔智达科技有限公司 | Multi-spectral photoelectric automatic recognition and tracking system |
| CN107851318B (en) * | 2015-08-18 | 2021-08-17 | 高通股份有限公司 | System and method for object tracking |
| CN107851318A (en) * | 2015-08-18 | 2018-03-27 | 高通股份有限公司 | System and method for Object tracking |
| CN106683120A (en) * | 2016-12-28 | 2017-05-17 | 杭州趣维科技有限公司 | Image processing method being able to track and cover dynamic sticker |
| CN106683120B (en) * | 2016-12-28 | 2019-12-13 | 杭州趣维科技有限公司 | image processing method for tracking and covering dynamic sticker |
| CN106952295A (en) * | 2017-03-17 | 2017-07-14 | 公安部第三研究所 | A kind of implementation method of the rotor wing unmanned aerial vehicle pursuit movement target of view-based access control model |
| CN110633731A (en) * | 2019-08-13 | 2019-12-31 | 杭州电子科技大学 | Single-stage anchor-frame-free target detection method based on staggered sensing convolution |
| CN110633731B (en) * | 2019-08-13 | 2022-02-25 | 杭州电子科技大学 | Single-stage anchor-frame-free target detection method based on staggered sensing convolution |
| CN115037869A (en) * | 2021-03-05 | 2022-09-09 | Oppo广东移动通信有限公司 | Automatic focusing method and device, electronic equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101226592A (en) | Method for tracing object based on component | |
| Hausler et al. | Multi-process fusion: Visual place recognition using multiple image processing methods | |
| Lieb et al. | Adaptive Road Following using Self-Supervised Learning and Reverse Optical Flow. | |
| Zhou et al. | Efficient road detection and tracking for unmanned aerial vehicle | |
| Akolkar et al. | Real-time high speed motion prediction using fast aperture-robust event-driven visual flow | |
| US8885876B2 (en) | Visual tracking system and method thereof | |
| Zhou et al. | Real time robust human detection and tracking system | |
| CN108596129A (en) | A kind of vehicle based on intelligent video analysis technology gets over line detecting method | |
| CN104992451A (en) | Improved target tracking method | |
| CN103927764B (en) | A kind of wireless vehicle tracking of combining target information and estimation | |
| CN103886325B (en) | Cyclic matrix video tracking method with partition | |
| CN110728697A (en) | Infrared dim target detection tracking method based on convolutional neural network | |
| CN104200495A (en) | Multi-target tracking method in video surveillance | |
| CN101976504B (en) | Multi-vehicle video tracking method based on color space information | |
| CN101800890A (en) | Multiple vehicle video tracking method in expressway monitoring scene | |
| CN104200485A (en) | Video-monitoring-oriented human body tracking method | |
| CN107169972B (en) | Non-cooperative target rapid contour tracking method | |
| CN107507223A (en) | Method for tracking target based on multi-characters clusterl matching under dynamic environment | |
| CN113269007A (en) | Target tracking device and method for road monitoring video | |
| Qing et al. | A novel particle filter implementation for a multiple-vehicle detection and tracking system using tail light segmentation | |
| Ma | An object tracking algorithm based on optical flow and temporal–spatial context | |
| CN106485283B (en) | A kind of particle filter pedestrian target tracking based on Online Boosting | |
| CN107886524B (en) | Method for identifying motion track of elevator door | |
| Du | CAMShift-Based Moving Object Tracking System | |
| CN108985216A (en) | A kind of pedestrian head detection method based on multiple logistic regression Fusion Features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20080723 |