CN108876820A - A kind of obstruction conditions based on average drifting move down object tracking method - Google Patents

A kind of obstruction conditions based on average drifting move down object tracking method Download PDF

Info

Publication number
CN108876820A
CN108876820A CN201810596691.4A CN201810596691A CN108876820A CN 108876820 A CN108876820 A CN 108876820A CN 201810596691 A CN201810596691 A CN 201810596691A CN 108876820 A CN108876820 A CN 108876820A
Authority
CN
China
Prior art keywords
target
kalman
region
model
formula
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.)
Granted
Application number
CN201810596691.4A
Other languages
Chinese (zh)
Other versions
CN108876820B (en
Inventor
蔡延光
赵豪
蔡颢
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN201810596691.4A priority Critical patent/CN108876820B/en
Publication of CN108876820A publication Critical patent/CN108876820A/en
Application granted granted Critical
Publication of CN108876820B publication Critical patent/CN108876820B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Multimedia (AREA)
  • Mathematical Physics (AREA)
  • Operations Research (AREA)
  • Probability & Statistics with Applications (AREA)
  • Algebra (AREA)
  • Evolutionary Biology (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a kind of obstruction conditions based on average drifting to move down object tracking method.Key step：1. obtaining the video frame images information in video data；2. selected tracking target, establishes kalman filter models, the initial parameter of initialized equations；3. establishing mean-shifted models, the color value of all pixels point in search window is counted, histogram is established and normalizes；4. calculating the predicted position of Kalman model output, as the initial iteration position of MeanShift algorithm, shadowing is carried out at the position of Kalman Prediction；5. if unobstructed, using predicted position as the initial iteration position of MeanShift algorithm, the optimal location of target is constantly iterated to calculate out according to mean shift algorithm, and update Kalman model parameter, the size of adaptive change window at optimal location, and the window size as next frame；6. if exporting the target position of prediction using predicted position as the observation position assumed block with this location updating Kalman model parameter.

Description

A kind of obstruction conditions based on average drifting move down object tracking method
Technical field
The present invention relates to field of image processings more particularly to a kind of obstruction conditions based on average drifting to move down moving-target and chase after Track method.
Background technique
The tracking of mobile target based on video, which refers to, carries out intelligence to the moving target of moving image present in video It detects and is tracked, while the associ-ated motion parameters index of moving target can be obtained.In specific practice, complicated ring The tracing algorithm that border moves down moving-target usually requires to detect segmentation, the track filtering of moving target with Intelligent target and to target Other aided algorithms such as the prediction of position, which are combined, carrys out effect of optimization, however because moving target local environment complexity and Bring is right as shadow of object, target be blocked, the variation on target morphology, brightness change etc. because of caused by illumination variation The requirement of algorithm has very big test, and the requirement of real-time in practical matter also target tracking problem is proposed it is very big It is required that.
Summary of the invention
The present invention in order to overcome at least one of the drawbacks of the prior art described above, provides a kind of blocking based on average drifting Under the conditions of mobile target tracking method, can be effectively solved and move down object tracking problem in obstruction conditions.
In order to solve the above technical problems, the technical solution adopted by the present invention is that：A kind of obstruction conditions based on average drifting Object tracking method is moved down, is included the following steps：
S1. the video frame images information in video data is obtained；
S2. initial search window is manually set in arbitrary frame, i.e., selected tracking target, and establish Kalman filter mould Type, while the initial parameter of initialized equations；
S3. mean-shifted models are established, the color value of all pixels point in search window is counted, establish histogram and normalizing Change, the model for obtaining target area describes to get the density probability function for having arrived target area；
S4. the predicted position of Kalman model output, the initial iteration position as MeanShift algorithm are calculated；And Shadowing is carried out at the position of Kalman Prediction, if not blocking, is executed step S5, is blocked if it exists, executes step S6；
If S5. not blocking, using the predicted position of Kalman model as the initial iteration position of MeanShift algorithm, The optimal location of target is constantly iterated to calculate out according to mean shift algorithm, and karr is updated using this optimal location as parameter Graceful model parameter, at optimal location, the size of adaptive change window, using this size as the window size of next frame；
S6. it blocks if it exists, using the predicted position of Kalman as the observation position assumed, and with this location updating karr Graceful model parameter exports the target position of prediction；
S7. judge whether video terminates, if so, step S8 is executed, if it is not, executing step S4；
Further, the S3 step specifically includes：
S31. object module is established, subregional selection is carried out to target using rectangular window, is extracted to feature When by region division be 11 sub-regions, 11 sub-regions constitute pentalpha structure, central area be regular pentagon, five jiaos Each angle of star is divided into a region；The color histogram for counting all subregion respectively, all subregion is counted respectively Color histogram result is described as the feature of entire target area；Assuming that the center of target area pixel is x0, zi(i=1, It 2 ..., n) indicates the coordinate position of each pixel in region, establishes color histogram, obtain the features that m is obtained by statistical color Value；The probability density q of object moduleu(u=1,2 ..., m) it is represented by：
In formula, k (x) is profile function, C quStandardization constant coefficient；U is the index of histogram；δ[b(zi- u)] it is to sentence Whether the brightness value in disconnected region template at pixel xi is located at u-th of section of histogram, and h is kernel function bandwidth, normalizes picture The position size of element；
S32. candidate region model describes, in t frame, regional center coordinate ft, with { ziI=1,2 ..., i, which is represented, to be waited The pixel of favored area, then the probability density of the model of candidate region be：
S33. measuring similarity, a kind of improved similarity judgment basis of proposition, expression are：
In formula, ρ (p, q) indicates the similarity between candidate target model and object module, ρi(p, q) (wherein i=1, 2 ..., the 11) similarity between different zones, Pasteur's coefficient description similarity is used herein；ωiRespectively represent each sub-district Domain weight coefficient indicates different contributions of the different zones to similarity between entire candidate family and object module；x0It represents The center-of-mass coordinate of central area, xiThe center-of-mass coordinate in other 10 regions is represented, the weight of different zones is by the region mass center Position inversely proportional distribution at a distance from the centroid position of central area.
Further, in the S4 step, the judgement blocked according to the following formula：
In formula, ρi(k) similarity in i-th each region, T are threshold value when representing kth frame；When the condition shown in above formula that meets When, it can be assumed that blocked at region 1, and general direction be since at region 1, similarly, other several regions Judgement also judged according to above formula.
5. a kind of obstruction conditions based on average drifting according to claim 2 move down object tracking method, It is characterized in that, the optimal location of target is iterated to calculate out in the S5 step constantly in order to make similar function ρ (p, q) most Greatly, Taylor expansion is used to the formula in step S33, obtains corresponding approximate expression：
Only have f that can change in above formula function ρ (p, q), so Section 2 is only analyzed, according to the opinion of Yizong Cheng Text can obtain following formula：
In formula, fkFor former target's center, fk+1Result central point as after average drifting calculates, if finally transported Dynamic distance is less than a certain threshold epsilon or has reached maximum number of iterations, it can assert that vector direction is towards two model face Colour contrast changes maximum direction movement, and position is the target optimal location of present frame after movement, then this specific center Point repeats this operation until terminating as the center of next frame iterative algorithm.
Compared with prior art, beneficial effect is：A kind of obstruction conditions based on average drifting provided by the invention move down Object tracking method is being established the model stage using a kind of improved subregion mode, i.e. pentalpha structure, is being compensated for The shortcomings that classical mean shift algorithm does not account for spatial character when establishing model to target, to the description of the model of target more added with Effect；The model can judge the directional information being blocked, to Kalman Prediction model when in face of blocking to a certain extent simultaneously There is certain correcting action, keeps it more adaptable to what is blocked.
Detailed description of the invention
Fig. 1 is the method for the present invention flow chart.
Fig. 2 is that subregional color histogram divides schematic diagram in step S31 of the present invention.
Fig. 3 is simulation result diagram of the embodiment of the present invention.
Specific embodiment
Attached drawing only for illustration, is not considered as limiting the invention；In order to better illustrate this embodiment, attached Scheme certain components to have omission, zoom in or out, does not represent the size of actual product；To those skilled in the art, The omitting of some known structures and their instructions in the attached drawings are understandable.Being given for example only property of positional relationship is described in attached drawing Illustrate, is not considered as limiting the invention.
Embodiment 1：
As shown in Figure 1, a kind of obstruction conditions based on average drifting move down object tracking method, include the following steps：
Step 1. obtains the video frame images information in video data.
Step 2. manually sets initial search window in arbitrary frame, i.e., selected tracking target, and establishes Kalman's filter Wave device model, and the initial parameter of initialized equations.It specifically includes：
S21. the state parameter for assuming mobile target is center and the speed of the area-of-interest at current time, with This is set in state variable in Kalman predictor and is set as Xk, observation is set as Zk, expression formula is distinguished as follows：
Xk=[xk,yk,vx,k,vy,k] (1)
Zk=[xk,yk] (2)
Wherein, xk,ykThat represent is the center position (particle position) of the rectangle comprising target area, vx,k, vy,kIt represents X, the movement speed size in y-axis；
S22. the state equation of Kalman predictor is：
Xk=Ak,k-1Xk-1+Wk-1 (3)
Wherein, Ak,k-1It indicates to be indicated with following matrix herein from the k-1 moment to the state-transition matrix at k moment：
In formula, Δ t indicates the time interval of continuous two frames picture, selects 1 frame here；Wk-1What it is for the k-1 moment is being The analogue noise added in system state, use irrelevant and mean value size for 0 normal white noise；
S23. observational equation is：
Zk=HkXk+Vk (5)
In formula, HkFor the observing matrix at k moment, VkFor the observation noise at k moment, using irrelevant normal state white noise The mean value of sound and Gaussian noise is set as 0.Observing matrix is as follows：
S24. the state renewal equation of Kalman predictor is：
The status predication equation of Kalman predictor is：
KkFor the gain matrix of Kalman predictor：
Wherein,For XkPrior probably estimation,For according to ZkIt is rightAmendment after Updated value, Wk-1And VkFor the noise of complementary relevant standardized normal distribution；QkAnd RkFor WkAnd VkCovariance matrix, specifically It is as follows：
Step 3. establishes mean-shifted models, counts the color value of all pixels point in search window, establishes histogram simultaneously Normalization, the model for obtaining target area at this time describe to have obtained target area (using color histogram as clarification of objective) Density probability function；It specifically includes：
The foundation of object module has used a kind of description method of improved target area feature, i.e., subregional target Block of pixels statistical color histogram still carries out the selection in region, as shown in Fig. 2, to feature using rectangular window to target Extraction when using the pentalpha region for being divided into 11 sub-regions, central area is a regular pentagon, the subregion point It Bian Hao not be 1~11, then count the color histogram in each region respectively, the color histogram result that each region is counted respectively As the feature description of entire target area, the description of the space characteristics distribution to target is improved.Assuming that target area pixel Center is x0, zi(i=1,2 ..., n) indicates the coordinate position of each pixel in region, establishes color histogram, obtains m by uniting The characteristic value that meter color obtains.The probability density q of object moduleu(u=1,2 ..., m) it is represented by：
In above formula, k (x) is profile function, C quStandardization constant coefficient；U is the index of histogram；δ[b(zi- u)] be Judge whether the brightness value in region template at pixel xi is located at u-th of section of histogram；
S32. candidate region model describes, in t frame, regional center coordinate ft, with { ziI=1,2 ..., i, which is represented, to be waited The pixel of favored area, then the probability density of the model of candidate region be：
S33. measuring similarity, similarity function are used to describe the similar journey between true model and target candidate model Degree, this paper similarity function will use Bhattacharyya coefficient, and formula is as follows：
A kind of improved similarity judgment basis proposed in this paper, expression are：
Wherein, ρ (p, q) indicates the similarity between candidate target model and object module, ρi(p, q) (wherein i=1, 2 ..., the 11) similarity between different zones, Pasteur's coefficient description similarity is used herein.ωiIt respectively represents different in Fig. 2 The weight coefficient in region indicates different contributions of the different zones to similarity between entire candidate family and object module, distribution Mode is carried out according to formula (20) (21), wherein xiRepresent the center-of-mass coordinate in 1~No. 10 region in 2 in figure, x0It represents 2 in figure In No. 11 regions center-of-mass coordinate, the weights of different zones by the region centroid position at a distance from the centroid position of central area at Inverse proportion distribution
Step 4. calculates the predicted position of Kalman model output, the initial iteration position as MeanShift algorithm；And Shadowing is carried out at the position of Kalman Prediction, if not blocking, is executed S5, is blocked if it exists, executes S6.
Wherein, the judgment basis blocked, shown in formula specific as follows：
Wherein, ρi(k) similarity in i-th each region when representing kth frame, T are threshold value, and value 0.8 specifically regards concrete condition Depending on.When meeting condition shown in formula (22), it can be assumed that being blocked at region 1, and general direction is from area Start at domain 1, is similarly adaptable to the judgement in remaining several region.
If step 5. is not blocked, using the predicted position of Kalman model as the initial iteration position of MeanShift algorithm It sets, the optimal location of target is constantly iterated to calculate out according to mean shift algorithm, and more using this optimal location as parameter New Kalman's model parameter, and at optimal location, the size of adaptive change window, using this size as the window of next frame Mouth size.
Constantly the optimal location of target is iterated to calculate out in order to keep similar function ρ (p, q) maximum, and formula (17) are transported With Taylor expansion, it is as follows to obtain corresponding approximate expression：
Only have f that can change in formula (23), so only analyzing Section 2, can be obtained according to the paper of Yizong Cheng Following formula：
In above formula, fkFor former target's center, fk+1Result central point as after average drifting calculates, Mean shift Algorithm is iterated calculating with formula (25), if the distance finally moved is less than a certain threshold epsilon or has reached greatest iteration Number, it can assert that vector direction is to change maximum direction movement towards two model color contrasts, position is after movement For the target optimal location of present frame, then this specific central point is treated as the center of next frame iterative algorithm, repeats this operation Until terminating, ε value is 0.5 in the present embodiment, and maximum number of iterations takes 20.
Step 6. is blocked if it exists, using the predicted position of Kalman as the observation position assumed, and with this location updating Kalman model parameter exports the target position of prediction.
Step 7. judges whether video terminates, if so, S8 is executed, if it is not, executing S4；
Step 8. stops, and terminates.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention Protection scope within.

Claims (4)

1. a kind of obstruction conditions based on average drifting move down object tracking method, which is characterized in that include the following steps：
S1. the video frame images information in video data is obtained；
S2. initial search window is manually set in arbitrary frame, i.e., selected tracking target, and kalman filter models are established, The initial parameter of initialized equations simultaneously；
S3. mean-shifted models are established, the color value of all pixels point in search window is counted, establishes histogram and normalize, The model for obtaining target area describes to get the density probability function for having arrived target area；
S4. the predicted position of Kalman model output, the initial iteration position as MeanShift algorithm are calculated；And in karr Shadowing is carried out at the position of graceful prediction, if not blocking, is executed step S5, is blocked if it exists, executes step S6；
If S5. not blocking, using the predicted position of Kalman model as the initial iteration position of MeanShift algorithm, foundation Mean shift algorithm constantly iterates to calculate out the optimal location of target, and Kalman's mould is updated using this optimal location as parameter Shape parameter, at optimal location, the size of adaptive change window, using this size as the window size of next frame；
S6. it blocks if it exists, using the predicted position of Kalman as the observation position assumed, and with this location updating Kalman's mould Shape parameter exports the target position of prediction；
S7. judge whether video terminates, if so, step S8 is executed, if it is not, executing step S4.
2. a kind of obstruction conditions based on average drifting according to claim 1 move down object tracking method, feature It is, the S3 step specifically includes：
S31. object module is established, subregional selection is carried out to target using rectangular window, it will when being extracted to feature Region division is 11 sub-regions, and 11 sub-regions constitute pentalpha structure, and central area is regular pentagon, five-pointed star Each angle is divided into a region；The color histogram of all subregion, the color that all subregion is counted respectively are counted respectively Histogram results are described as the feature of entire target area；Assuming that the center of target area pixel is x0, zi(i=1,2 ..., N) coordinate position for indicating each pixel in region, establishes color histogram, obtains the m characteristic values obtained by statistical color；Mesh Mark the probability density q of modelu(u=1,2 ..., m) it is represented by：
In formula, k (x) is profile function, C quStandardization constant coefficient；U is the index of histogram；δ[b(zi- u)] it is to judge area Pixel x in the template of domainiWhether the brightness value at place is located at u-th of section of histogram, and h is kernel function bandwidth, normalizes pixel Position size；
S32. candidate region model describes, in t frame, regional center coordinate ft, with { ziI=1,2 ..., i represents candidate regions The pixel in domain, then the probability density of the model of candidate region be：
S33. measuring similarity, a kind of improved similarity judgment basis of proposition, expression are：
In formula, ρ (p, q) indicates the similarity between candidate target model and object module, ρi(p, q) (wherein i=1,2 ..., 11) similarity between different zones uses Pasteur's coefficient description similarity herein；ωiRespectively represent each sub-regions weight Coefficient indicates different contributions of the different zones to similarity between entire candidate family and object module；x0Represent center The center-of-mass coordinate in domain, xiRepresent the center-of-mass coordinate in other 10 regions, the weights of different zones by the region centroid position with The inversely proportional distribution of the distance of central area centroid position.
3. a kind of obstruction conditions based on average drifting according to claim 2 move down object tracking method, feature It is, in the S4 step, the judgement blocked according to the following formula：
In formula, ρi(k) similarity in i-th each region, T are threshold value when representing kth frame；It, can be with when meeting condition shown in above formula Assert region 1 at blocked, and general direction be since at region 1, similarly, the judgement in other several regions Judged according to above formula.
4. a kind of obstruction conditions based on average drifting according to claim 2 move down object tracking method, feature It is, the optimal location of target is iterated to calculate out in the S5 step constantly in order to keep similar function ρ (p, q) maximum, Taylor expansion is used to the formula in step S33, obtains corresponding approximate expression：
Only have f that can change in above formula function ρ (p, q), so Section 2 is only analyzed, it can according to the paper of Yizong Cheng Obtain following formula：
In formula, fkFor former target's center, fk+1Result central point as after average drifting calculates, if finally moved away from From being less than a certain threshold epsilon or reached maximum number of iterations, it can assert that vector direction is towards two model color contrasts Change that maximum direction is mobile, it is mobile after position be present frame target optimal location, then this specific central point as The center of next frame iterative algorithm repeats this operation until terminating.
CN201810596691.4A 2018-06-11 2018-06-11 Moving target tracking method under shielding condition based on mean shift Active CN108876820B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810596691.4A CN108876820B (en) 2018-06-11 2018-06-11 Moving target tracking method under shielding condition based on mean shift

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810596691.4A CN108876820B (en) 2018-06-11 2018-06-11 Moving target tracking method under shielding condition based on mean shift

Publications (2)

Publication Number Publication Date
CN108876820A true CN108876820A (en) 2018-11-23
CN108876820B CN108876820B (en) 2022-01-25

Family

ID=64337765

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810596691.4A Active CN108876820B (en) 2018-06-11 2018-06-11 Moving target tracking method under shielding condition based on mean shift

Country Status (1)

Country Link
CN (1) CN108876820B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109949340A (en) * 2019-03-04 2019-06-28 湖北三江航天万峰科技发展有限公司 Target scale adaptive tracking method based on OpenCV
CN110233667A (en) * 2019-06-05 2019-09-13 华南理工大学 VLC dynamic positioning method and system based on average drifting and Unscented kalman filtering
CN110458862A (en) * 2019-05-22 2019-11-15 西安邮电大学 A kind of motion target tracking method blocked under background
CN110517291A (en) * 2019-08-27 2019-11-29 南京邮电大学 A kind of road vehicle tracking based on multiple feature spaces fusion
CN111277745A (en) * 2018-12-04 2020-06-12 北京奇虎科技有限公司 Target person tracking method and device, electronic equipment and readable storage medium
CN112070794A (en) * 2020-08-20 2020-12-11 成都恒创新星科技有限公司 Multi-object tracking method based on dynamic auxiliary target
CN115471139A (en) * 2022-10-31 2022-12-13 北京奥邦体育赛事评估有限责任公司 Large-scale crowd sports event comprehensive evaluation system based on image recognition technology

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101324956A (en) * 2008-07-10 2008-12-17 上海交通大学 Method for tracking anti-shield movement object based on average value wander
US20160140397A1 (en) * 2012-01-17 2016-05-19 Avigilon Fortress Corporation System and method for video content analysis using depth sensing
US20170061239A1 (en) * 2015-05-22 2017-03-02 International Business Machines Corporation Real-time object analysis with occlusion handling

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101324956A (en) * 2008-07-10 2008-12-17 上海交通大学 Method for tracking anti-shield movement object based on average value wander
US20160140397A1 (en) * 2012-01-17 2016-05-19 Avigilon Fortress Corporation System and method for video content analysis using depth sensing
US20170061239A1 (en) * 2015-05-22 2017-03-02 International Business Machines Corporation Real-time object analysis with occlusion handling

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
赵豪等: "基于卡尔曼的均值漂移抗遮挡移动目标追踪算法", 《电子世界》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111277745A (en) * 2018-12-04 2020-06-12 北京奇虎科技有限公司 Target person tracking method and device, electronic equipment and readable storage medium
CN111277745B (en) * 2018-12-04 2023-12-05 北京奇虎科技有限公司 Target person tracking method and device, electronic equipment and readable storage medium
CN109949340A (en) * 2019-03-04 2019-06-28 湖北三江航天万峰科技发展有限公司 Target scale adaptive tracking method based on OpenCV
CN110458862A (en) * 2019-05-22 2019-11-15 西安邮电大学 A kind of motion target tracking method blocked under background
CN110233667A (en) * 2019-06-05 2019-09-13 华南理工大学 VLC dynamic positioning method and system based on average drifting and Unscented kalman filtering
CN110517291A (en) * 2019-08-27 2019-11-29 南京邮电大学 A kind of road vehicle tracking based on multiple feature spaces fusion
CN112070794A (en) * 2020-08-20 2020-12-11 成都恒创新星科技有限公司 Multi-object tracking method based on dynamic auxiliary target
CN115471139A (en) * 2022-10-31 2022-12-13 北京奥邦体育赛事评估有限责任公司 Large-scale crowd sports event comprehensive evaluation system based on image recognition technology
CN115471139B (en) * 2022-10-31 2023-02-10 北京奥邦体育赛事评估有限责任公司 Large-scale crowd sports event comprehensive evaluation system based on image recognition technology

Also Published As

Publication number Publication date
CN108876820B (en) 2022-01-25

Similar Documents

Publication Publication Date Title
CN108876820A (en) A kind of obstruction conditions based on average drifting move down object tracking method
CN106846359B (en) Moving target rapid detection method based on video sequence
CN110517288B (en) Real-time target detection tracking method based on panoramic multi-path 4k video images
CN104200485B (en) Video-monitoring-oriented human body tracking method
CN106997597B (en) It is a kind of based on have supervision conspicuousness detection method for tracking target
CN104504394B (en) A kind of intensive Population size estimation method and system based on multi-feature fusion
CN105404847B (en) A kind of residue real-time detection method
CN101315701B (en) Movement destination image partition method
CN110688987A (en) Pedestrian position detection and tracking method and system
CN105184812B (en) A kind of pedestrian based on target following hovers detection method
CN109919053A (en) A kind of deep learning vehicle parking detection method based on monitor video
CN109063559A (en) A kind of pedestrian detection method returned based on improvement region
CN106204586B (en) A kind of moving target detecting method under complex scene based on tracking
CN105279769B (en) A kind of level particle filter tracking method for combining multiple features
CN101916446A (en) Gray level target tracking algorithm based on marginal information and mean shift
CN106127812B (en) A kind of passenger flow statistical method of the non-gate area in passenger station based on video monitoring
CN104167006B (en) Gesture tracking method of any hand shape
CN102142085B (en) Robust tracking method for moving flame target in forest region monitoring video
CN106709938B (en) Based on the multi-target tracking method for improving TLD
CN110415268A (en) A kind of moving region foreground image algorithm combined based on background differential technique and frame difference method
CN107767404A (en) A kind of remote sensing images sequence moving target detection method based on improvement ViBe background models
CN108596920A (en) A kind of Target Segmentation method and device based on coloured image
CN106327488A (en) Adaptive foreground detection method and detection device
CN104036526A (en) Gray target tracking method based on self-adaptive window
CN112258525B (en) Image abundance statistics and population identification algorithm based on bird high-frame frequency sequence

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
GR01 Patent grant
GR01 Patent grant