CN101692285A - Real-time color video tracking algorithm for supercomplex totally described color images - Google Patents
Real-time color video tracking algorithm for supercomplex totally described color images Download PDFInfo
- Publication number
- CN101692285A CN101692285A CN200910194491A CN200910194491A CN101692285A CN 101692285 A CN101692285 A CN 101692285A CN 200910194491 A CN200910194491 A CN 200910194491A CN 200910194491 A CN200910194491 A CN 200910194491A CN 101692285 A CN101692285 A CN 101692285A
- Authority
- CN
- China
- Prior art keywords
- target
- supercomplex
- tracking
- coloured image
- template
- 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
The invention belongs to the technical field of video image processing, and in particular relates to a related real-time color video-based new method for tracking a target. In the method, the target is tracked in a mode of searching a local color image, and correlation factors of supercomplex totally described color image quality are utilized as characteristic parameters. The characteristic parameters describe the similarity among the color images so as to represent the level of the similarity between the tracked target and a target template according to the values of the correlation factors of the color image quality; and at the same time, the color information of the color video can be fully and effectively utilized so as to achieve the aim of accurately tracking the target. Experimental results show that the characteristic parameters describing the target and a corresponding tracking algorithm can be used to effectively track the target in the color video.
Description
Technical field
The invention belongs to technical field of video image processing, be specially a kind of characteristic parameter that adopts the mass parameter of the whole description of supercomplex coloured image as target following, and carry out the method for real-time color video frequency object tracking.
Background technology
Motion target detection and tracking in the sequence image are analyzed the image sequence of taking by being meant, detect moving target and calculate the position of target on every two field picture, provide the correlation parameter of relevant video object motion and the method for rule then.This problem is one of main contents of computer vision research.Computer vision is a multidisciplinary cross subject, relates to Flame Image Process, computer graphics, pattern-recognition, artificial intelligence, neural network, computing machine, psychology, physiology, physics and mathematics etc.
The video tracking technology is requisite gordian technique in the computer vision, and it all is widely used aspect many in robot visual guidance, military visual guidance, safety monitoring, traffic control, medical diagnosis, video compress and meteorologic analysis etc.As military aspect, Imaging Guidance, military surveillance and the supervision etc. of weapon successfully have been applied to.Early stage TV and infrared tracker all adopt single mode of operation, must realize with hardware fully.Present tracking can computing machine be the basis, adopts image processing and pattern recognition, utilizes programmed control to realize multiple function.The multimode tracker has been used for TV and infrared imaging system, as the guidance system of guided missiles such as the tank knockout person of the U.S. and Haier's method.In recent years, artificial intelligence is applied in the video tracking, has improved the adaptive ability of system effectively.Civilian aspect as vision monitoring, has been widely used in the each side of social life.Video tracking can be applicable to the guard monitor of community and critical facility; Be used for intelligent transportation system and carry out the real-time detection and the tracking of vehicle, can obtain the many valuable traffic flow parameters of vehicle flowrate, vehicle, the speed of a motor vehicle, vehicle density or the like, simultaneously can also the detection accident or emergency situations such as fault.The FB(flow block) of total system as shown in Figure 1.
A most important problem is exactly how to represent tracked target in the video tracking technology.Existing method has the method [1] based on bank of filters, based on the method [2] of subspace with based on the relevant method of template [3].Wherein easy owing to realizing based on the relevant method of template, the efficient height of algorithm is used in the application of reality widely.
Main thought based on the relevant video frequency object tracking algorithm of template is, set up the foundation of the template of an expression target or examined object in advance as identification and definite target location, compare with each the sub regions image in To Template and the realtime graphic, find out the position of a number of sub images the most similar, just think the position of current goal with To Template.But, traditional video frequency object tracking algorithm of being correlated with based on template has several significant disadvantages.At first, strong inadequately based on the robustness of the relevant video frequency object tracking algorithm of template, some illumination variation of environment will seriously influence the effect of tracking.Secondly, only be suitable for black and white video based on relevant video tracking algorithm and follow the tracks of, because it does not consider the color information in the color video, the performance in color video is followed the tracks of is undesirable.
Summary of the invention
The object of the present invention is to provide a kind of new method based on relevant real-time color video frequency object tracking, propose with the whole mass parameter of coloured image of describing of supercomplex as detecting the clarification of objective parameter, thereby take all factors into consideration red (R) of coloured image on the whole, green (G), therefore the full detail of blue (B) three primary colours can reflect the similarity of coloured image better.
1843, the Hamilton proposed a useful plural number and has been referred to as supercomplex, also claims hypercomplex number.Supercomplex can be seen the popularization that pluralizes, and it comprises a real component and three imaginary part components.The point in supercomplex space can be expressed as
q(n)=q
0(n)+iq
1(n)+jq
2(n)+kq
3(n) (1)
And definition:
If wish to describe coloured image R, G, the contact of the inherence of B three colouring components, the pure supercomplex that the supercomplex of formula (1) can be carried out following no real part as a vector integral body with three colouring components of coloured image is described [4]:
f(m,n)=R(m,n)i+G(m,n)j+B(m,n)k (3)
(m, n), (m, n), (m n) is illustrated respectively in coloured image coordinate (m, n) locational R, G, the numerical value of B three colouring components to B to G to R in the formula.Be provided with two width of cloth coloured image X and Y, X={x
i| i=1, Λ M}, Y={y
i| i=1, Λ M}, wherein x
mAnd y
mBe the coloured image pixel of representing with pure supercomplex, that is:
x
m=x
r(m)i+x
g(m)j+x
b(m)k (4)
y
m=y
r(m)i+y
g(m)j+y
b(m)k
X in the formula
r(m), x
g(m), x
b(m) and y
r(m), y
g(m), y
b(m) be respectively the R of X and Y two width of cloth coloured images, the component of G and B.
To two width of cloth coloured image X and the general color image quality index (UCQI) of Y definition is [6]:
In the formula
| μ
xμ
y| be supercomplex average μ
xAnd μ
yThe mould value of product, supercomplex σ
XyPolar form be σ
Xy=| σ
Xy| e
-μ θ, and μ
x, μ
y, σ
xAnd σ
yBe pure supercomplex, according to hypercomplex multiplying rule, pure hypercomplex square is real number, its result equal pure onlap the digital-to-analogue value square.The quality index of formula (5) definition has been considered color distortion, associated loss, brightness and contrast's distortion, and R, G, the combination of distortion between B three looks.For these distortions are described, we can be rewritten as following form with formula (5)
Formula considered in (6) | PQ|=|P||Q|
We know that hypercomplex mould value represented the monochrome information of coloured image, so in the formula (6) first in the difference in brightness of measuring coloured image X and Y, second evaluation be the similarity of the contrast of coloured image X and Y, and the 3rd related coefficient of representing coloured image X and Y, its mould value is estimated the color distortion [6] between this two width of cloth coloured image by its argument θ in the linear dependence of measuring coloured image X and Y, and do not have color distortion [6] between θ=0 expression, two width of cloth coloured images.In brief, the mould value of formula (6) has reflected the information of two width of cloth coloured image spaces and structural similarity; And e
-μ θReflected the variation of two width of cloth coloured image color informations, be equivalent to two width of cloth coloured images are calculated the result that simple crosscorrelation obtains.Need lay special stress on to be, because
Therefore the coloured image X of formula (6) measurement and the related coefficient of Y have been considered R, G, the inner link of B three colouring components.Be the R that dot product in the formula (7) and multiplication cross are expressed, G, the inner link of B three colouring components.
The use image quality parameter that the present invention proposes is as the characteristic parameter that detects, and flow process is shown in Figure 2.
The flow process of the target tracking algorism of topography's search is:
(1) choose the target and the corresponding To Template of tracking in first two field picture, To Template is for just comprising the rectangle frame of target.
(2) scope of selected search in the current picture frame that carries out target following: the center of choosing with To Template is the center, with the multiple of To Template length and width (for example 2 times, 3 times ...., etc.) for a rectangular area of length and width be the Local Search scope.
(3) use To Template that the subimage of sizes such as each and To Template in the hunting zone is carried out the color image quality CALCULATION OF PARAMETERS, i.e. calculating formula (6).When its numerical value reaches maximum, think that then target following is successful.
(4) threshold value template renewal.Investigate the numerical value of the color image quality parameter of current tracking success, promptly the numerical value of formula (6) greater than some threshold values (a large amount of experiments show that 0.9 is an appropriate threshold), is updated to To Template with the subimage of correspondence as if it; If it less than this threshold value, shows that the partial occlusion situation has appearred in the target of this secondary tracking, continue to use original To Template.
Technique effect
In different application scenarioss, carried out a large amount of experiments.Fig. 3 to Fig. 4 for wherein two follow the tracks of examples.Can see that in different scenes at different tracking targets, this algorithm all is highly effective.
In addition, for the tracking factor of the use of verifying us characteristics to the coloured image sensitivity, Fig. 5 has provided the experimental example of a vehicle tracking.Tracking target is blue dolly, and its jamming target is the red dolly of the identical same model of profile.A red dolly identical with blue dolly shape is arranged again in the k two field picture.
And if the supercomplex color image quality correlation factor simple crosscorrelation coupling that we adopt this paper to propose is carried out target following, coloured image is expressed as the supercomplex form, the curved surface of then normalized supercomplex phase place simple crosscorrelation as shown in Figure 6, maximal value is blue dolly position, tracking target is correct.And as can be seen, curved surface is more smooth, and the peaked place of curved surface is more sharp-pointed.And, has only a peak on the surface chart of our correlation factor, just, our colored supercomplex correlation factor can not be subjected to the influence of the object of different colours, the result who is target following can not lose tracked target because of the object interference of identical other color of same model of profile, and this interference is a unvanquishable difficulty when only adopting black white image to follow the tracks of.
Description of drawings:
Fig. 1 is for following the tracks of example 1, and wherein (a) is the 1st frame, (b) is the 30th frame, (c) is the 60th frame, (d) is the 90th frame.
Fig. 2 is for following the tracks of example 2, and wherein (a) is the 1st frame, (b) is the 40th frame, (c) is the 80th frame, (d) is the 120th frame.
Fig. 3 is the colored target tracking experiment, and wherein (a) is tracking target, (b) is k two field picture to be matched, is the region of search in the square frame.
Fig. 4 is correlation factor figure.
List of references
[1].Z.Wang,A.C.Bovik,H.R.Sheikh,E.P.Simoncelli.“Image?quality?assessment:From?error?measurement?to?structural?similarity”.IEEE?Trans.Image?Process,vol.13,no.4,pp.600-612,Apr.2004,
[2].A.D.Jepson,D.J.Fleet,and?T.F.El-Maraghi,“Robust?online?appearancemodels?for?visual?tracking.”IEEE?Trans.Pattern?Analysis?and?MachineIntelligence,vol.25,no.10,pp.1296-1311,Oct.2003
[3].J.Lim,D.Ross,R.S.Lin,and?M.H.Yang,“Incremental?learning?for?visualtracking,”Advances?in?neural?information?processing?systems,vol.17,pp.793-800,2005.
[4].D.A.Montera,S.K.Rogers,D.W.Ruck,and?M.E.Oxley,“Object?tracking?throughadaptive?correlation,”Proceedings?of?SPIE,1959.
[5].C.E.Moxey,S.J.Sangwine,T.A.Ell.“Hypercomplex?correlation?techniquesfor?vector?images,”IEEE?Trans?on?Signal?Processing,vol.51,no.7?pp.1941-1953,July?2003.
[6]. Hao Mingfei, " research of whole splicing of colorful image hypercomplex number and quality evaluation thereof " Fudan University's Master of Science degree paper, in May, 2007.
Claims (4)
1. the new method of a real-time color video frequency object tracking is characterized in that using " the whole quality correlation factor of describing coloured image of supercomplex " as tracking color video clarification of objective parameter, and the tracking of Local Search and update strategy.
2. the new method of real-time color video frequency object tracking according to claim 1 is characterized in that being applied to the characteristic parameter of target following---the whole mass parameter of describing coloured image of supercomplex.The Global Information of coloured image has been considered in this parametric synthesis, can well reflect the difference of two width of cloth coloured images for human eye.Therefore can reach the target following ability of human eye.
3. according to the new method of claim 1,2 described real-time color video frequency object trackings, it is characterized in that application tracking clarification of objective parameter---the mass parameter of coloured image is carried out target following by the method for Local Search: the center with To Template is the center, twice with the To Template length and width is length and width, by calculating this zone color image quality parameter this zone is searched for.When the image quality parameter value that provides reaches maximum, the target following success.
4. according to the method for claim 1,2,3 described real-time video target trackings, it is characterized in that the characteristic parameter that application target is followed the tracks of---after the whole quality correlation factor of describing coloured image of supercomplex is carried out the target following success by the method for Local Search, carry out the threshold value template renewal: the quality correlation factor numerical value of describing coloured image when supercomplex is whole is template greater than some threshold values with the current target update that traces into; When its less than some threshold values, illustrated that partial occlusion takes place, not new template more.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200910194491A CN101692285A (en) | 2009-08-24 | 2009-08-24 | Real-time color video tracking algorithm for supercomplex totally described color images |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200910194491A CN101692285A (en) | 2009-08-24 | 2009-08-24 | Real-time color video tracking algorithm for supercomplex totally described color images |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101692285A true CN101692285A (en) | 2010-04-07 |
Family
ID=42080968
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN200910194491A Pending CN101692285A (en) | 2009-08-24 | 2009-08-24 | Real-time color video tracking algorithm for supercomplex totally described color images |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101692285A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104751823A (en) * | 2015-04-16 | 2015-07-01 | 曲阜师范大学 | Characterizing method applicable to display screens of mobile terminals |
CN105796053A (en) * | 2015-02-15 | 2016-07-27 | 执鼎医疗科技(杭州)有限公司 | Method for measuring dynamic contrast ratio and estimating transverse flow with OCT |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1913645A (en) * | 2006-08-17 | 2007-02-14 | 复旦大学 | Color image quality estimating method based on super-complex |
CN101175203A (en) * | 2007-11-19 | 2008-05-07 | 上海龙东光电子有限公司 | Real-time video target tracking method |
-
2009
- 2009-08-24 CN CN200910194491A patent/CN101692285A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1913645A (en) * | 2006-08-17 | 2007-02-14 | 复旦大学 | Color image quality estimating method based on super-complex |
CN101175203A (en) * | 2007-11-19 | 2008-05-07 | 上海龙东光电子有限公司 | Real-time video target tracking method |
Non-Patent Citations (3)
Title |
---|
江淑红 等: "快速超复数傅氏变换和超复数互相关的新算法及应用", 《电子学报》 * |
江淑红: "彩色图像的超复数整体处理理论及应用研究", 《中国博士学位论文全文数据库》 * |
郝明非: "彩色图像超复数整体拼接及其质量评估的研究", 《万方学位论文》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105796053A (en) * | 2015-02-15 | 2016-07-27 | 执鼎医疗科技(杭州)有限公司 | Method for measuring dynamic contrast ratio and estimating transverse flow with OCT |
CN104751823A (en) * | 2015-04-16 | 2015-07-01 | 曲阜师范大学 | Characterizing method applicable to display screens of mobile terminals |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US12072705B2 (en) | Intelligent decision-making method and system for unmanned surface vehicle | |
CN114202696B (en) | SAR target detection method and device based on context vision and storage medium | |
Dai et al. | Multi-task faster R-CNN for nighttime pedestrian detection and distance estimation | |
CN102881022B (en) | Concealed-target tracking method based on on-line learning | |
CN102915650B (en) | Based on the bridge waters ship navigation safe early warning equipment of intersection photography | |
CN106354816A (en) | Video image processing method and video image processing device | |
CN110602449A (en) | Intelligent construction safety monitoring system method in large scene based on vision | |
Qu et al. | Moving vehicle detection with convolutional networks in UAV videos | |
Hu et al. | Face Detection based on SSD and CamShift | |
CN114972418A (en) | Maneuvering multi-target tracking method based on combination of nuclear adaptive filtering and YOLOX detection | |
CN106127812A (en) | A kind of passenger flow statistical method of non-gate area, passenger station based on video monitoring | |
CN103617630A (en) | Infrared image dividing method and system for power system equipment based on wavelet analysis | |
CN104318198A (en) | Identification method and device suitable for substation robot patrolling | |
CN102254394A (en) | Antitheft monitoring method for poles and towers in power transmission line based on video difference analysis | |
CN107833239A (en) | A kind of searching of optimal matching method for tracking target based on weighted model constraint | |
CN102289822A (en) | Method for tracking moving target collaboratively by multiple cameras | |
CN101175203A (en) | Real-time video target tracking method | |
CN116740988A (en) | Full-time bridge self-adaptive ship collision prevention early warning method of visible light and thermal imaging camera | |
CN101692285A (en) | Real-time color video tracking algorithm for supercomplex totally described color images | |
Jie et al. | Llformer: An efficient and real-time lidar lane detection method based on transformer | |
CN103996207A (en) | Object tracking method | |
CN103778641B (en) | Method for tracking target based on Wavelet Descriptor | |
CN103745486A (en) | Method for eliminating noise interference by using moving track of object | |
CN116030404A (en) | Artificial intelligence-based construction and safety monitoring method for electronic warning fence of operation area | |
Sun et al. | The study on intelligent vehicle collision-avoidance system with vision perception and fuzzy decision making |
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 |
Application publication date: 20100407 |