CN108133487A - The trans-regional single human body attitude target detection extracting method of video - Google Patents
The trans-regional single human body attitude target detection extracting method of video Download PDFInfo
- Publication number
- CN108133487A CN108133487A CN201711255086.2A CN201711255086A CN108133487A CN 108133487 A CN108133487 A CN 108133487A CN 201711255086 A CN201711255086 A CN 201711255086A CN 108133487 A CN108133487 A CN 108133487A
- Authority
- CN
- China
- Prior art keywords
- image
- formula
- background
- matrix
- video
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06T7/20—Analysis of motion
- G06T7/292—Multi-camera tracking
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of trans-regional single human body attitude target detection extracting method of video, the application proposes the moving object detection optimization algorithm based on compressed sensing for the trans-regional single human body attitude Objective extraction problem of multiple-camera human body sport video.The algorithm is more to newly arrive to realize while reconstruct by the alternating iteration of video sequence, background and prospect.Motion artifacts, sunlight variation can effectively be inhibited, realize the robustness of target detection.Present image is compared using the background image of optimization solution, using given threshold, single human body attitude target is extracted using binary image.
Description
Technical field
The present invention relates to a kind of trans-regional single human body attitude target detection extracting methods of video.
Background technology
Computer vision field, on the basis of advanced visual target tracking algorithm is studied, long-term follow to single goal,
The problems such as target tracker, target detection track device, with reference to new academic frontier knowledge, it is proposed that innovation intelligent algorithm is finally applied to
Trans-regional human body video frequency object tracking intelligent monitor system carries out single goal long-term follow, positioning and identification, has preferable skill
Art performance.
The long-term follow of single goal refers to after using a bounding box that single tracked target is determined, in continuous frame
The interior process that recognition and tracking is detected to the target.During tracking, tracking need to overcome environment and by
The continually changing interference of target is tracked, realizes long-term target following.The interference for generally requiring solution includes ambient lighting
Variation, noise, the movement of video sensor, occlusion issue, target leaves scene and target is again introduced into scene.Target is left
Scene and target are again introduced into the detection of scene and lasting tracking.The application mainly target is left scene and target again into
A kind of optimization method for entering the Detection and Extraction of scene and proposing, so as to improve the accuracy and reliability of target acquisition.
Invention content
For overcome the deficiencies in the prior art, the present invention provides a kind of video trans-regional single human body attitude target detection extraction
Method.
The technical solution adopted by the present invention to solve the technical problems is：
It is identical with claim.
The beneficial effects of the invention are as follows：The application is based on the trans-regional single human body attitude target of multiple-camera human body sport video
Extraction problem proposes one based on compressed sensing moving object detection optimization algorithm, to video while video sequence acquires
Sequence image is compressed, and improves video sequence calculating speed in detection process of moving target, reduces the space of storage.For
The moving object detection Optimized model of proposition has obtained background, the optimization solution of foreground image.Carried algorithm meets the detection of target mark
Reconstruction accuracy, prospect is good with background segment effect, can effectively inhibit motion artifacts, sunlight variation, realize moving object detection
Robustness.For background subtraction method algorithm real-time it is poor the shortcomings that, propose based on optimization solution background subtraction method, stablized
Prospect, background image.Algorithm is carried by experiment show and fully meets the mostly camera shooting of intelligent visual surveillance system high definition
The trans-regional target detection of head human motion, tracking, identification and adaptation Saeng requirements, and it has been experimentally confirmed the correctness of the algorithm
And feasibility.
It can be seen that, when a humanbody moving object occurs, algorithm can detect human motion quickly in an experiment
Target, when target occurs again, algorithm can quickly detect the humanbody moving object of appearance, line trace of going forward side by side.It can by figure
To find out that this algorithm can be detected correctly and continue tracking former single human body attitude target is extracted with identification.
This experiment show puies forward algorithm target mark detection reconstruction accuracy, prospect and background segment, can effectively press down
Motion artifacts processed realize the robustness of target detection.Itd is proposed algorithm can accurately detect humanbody moving object, ideal to examine
Survey, tracking and recognition effect solve the problems, such as to extract single human body attitude target in intelligent monitor system.
Description of the drawings
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1 is the principle of the present invention block diagram.
Specific embodiment
With reference to Fig. 1, the invention discloses a kind of trans-regional single human body attitude target detection extracting method of video,
Include the following steps：
One, the multi-frame video figure of the trans-regional moving target of multiple-camera human body sport video is acquired from intelligent monitor system
Picture establishes moving object detection Optimized model：
Wherein, X ∈ Rmn×KFor video sequence；B and F is the background and prospect of video, and prospect is the moving target to be looked for；|
| | | is nuclear norm, the norm calculation be singular values of a matrix and, the effect that nuclear norm is selected in formula (1) is constraint background
Rank of matrix, | | | |1For 1 norm,
Two, compressed sensing processing is carried out to multi-frame video image, video sequence is carried out while video sequence acquires
Compression,
Video foreground and background reconstruction are done directly from compression observing matrix A, and with robustness to realize, it can
Compression observing matrix A is obtained by the three-dimensional circular method of sampling, by compression sampling expression formulaIt is divided into two samplings
Step:Random convolution process Rk=Cxk；Random sampling procedure SkPiRk=ak, (k=1,2 .., K) enables W=0, by above-mentioned two
Sampling step is dissolved into optimization object function, and moving object detection Optimized model is as follows：
R=CX, SkPiRk=ak(k=1,2 ..K) (2)
Wherein, αiFor weighted value (i=1.2,3), DiFor difference operator, W is motion artifacts, and C is circular matrix, xkTo regard
Frequency frame, R are cycle observing matrix, and the initial value of matrix W is set as 0 during (2) formula of solution, when solving the excellent of X, B and F
Motion artifacts matrix W is calculated again after changing estimated value, the optimization problem of X, B and F is then solved, when solving-optimizing problem obtains X, B
And motion artifacts W can be distinguished during F, this ensure that the robustness of B and F segmentations,
In order to solve formula (2), it is more to newly arrive realization together by the alternating iteration of video sequence, background and prospect to carry algorithm
When reconstruct, first image reconstruction procedure be constantly update video sequence X reconstruction accuracy；Then carrying out image segmentation process is
The segmentation for realization prospect and the background of more being newly arrived by each video, in image reconstruction procedure, by solving following moving targets
Inspection optimization model realizes the reconstruct of video sequence,
Optimization problem involved by formula (3) can be solved by augmented vector approach, in solution procedure, be needed
The first Lagrange's equation of structure formula (3),
SkPiRk=ak(k=1,2 ..., K) (4)
In Lagrange's equation is built, μi=(i=1,2,3) and ρ is Lagrange multiplier matrix, passes through Lagrange
The structure of equation by the required majorization of solutions problem of formula (3), passes through the Lagrangian and glug in iteration newer (4)
Bright day multiplier is solved, and Lagrangian and Lagrange multiplier renewal process are
Utilize D in alternately newer (5)iX (i=1.2.3) is video sequence matrix of variables, and R and the used alternatings of X are more
New specific implementation process is:
Formula (6) is solved by the update of (9) formula：
Wherein:Function Sa() is threshold value iteration operator, and for scalar x, which is specifically defined as
Sa(x)=sign (x) max x | and-a, 0 } (10)
Wherein:A is iteration threshold, and when processing array threshold value iteration updates, we carry out threshold using formula (10) to matrix
It is worth iteration update, after matrix update is completed, output result updates matrix for threshold value, can pass through following step to formula (7) solution
It completes:
Wherein:Rk=(k=1,2 ... it is n) the kth row in cycle observing matrix R, PiCSaIt observes and replacing for stochastical sampling
Matrix, for representing cycle observing matrix RkIn which element needs be iterated update, PiCSaIt is by stochastical sampling matrix
SiAnd generate, the middle X updates of formula (8) are realized by solving least square problem,
During display foreground background segment, image reconstruction update result X fixed firstk+1, then by solving following formula
(13) optimization problem to reconstruct prospect and background simultaneously,
It is by the Lagrangian that formula (13) is established
Wherein:For H Lagrange multiplier matrixes, ＜, ＞ matrix inner products equally use augmentation Lagrangian Arithmetic pair
Formula (13) is solved, and process is as follows:
HK+1=HK+λ(XK+1-BK+1-FK+1) (16)
Formula (15) equally solves it using alternately update, display foreground context update process is as follows,
Three, it is more newly arrived by the alternating iteration of video sequence, background and prospect and realizes while reconstruct, asked using restructing algorithm
The detailed process of solution formula (2) is as follows:
Input:C,Pi,Sk,A
Output:X,B,F
Initial value is set as:
B0=F0=R=H0=ρ0=0
When meeting condition:||Xk+1-Xk||F/Xk||F＞ 10-6It continues cycling through
Image reconstruction procedure:Update:DiX,R
Update:X
Display foreground background segment process is as follows:
Bk+1=Di(Xk+1-Fk-Bk+Hk)
Fk+1=Sk(Xk+1-Bk+1-Fk+Hk)
K=k+1
End loop,
Four, background image is compared using the present image picture of target detection optimization solution, obtains frame difference image：
We obtain background image stability height, foreground image by solving video sequence, background and prospect optimization problem
Detection result is good as a result, compare to obtain foreground moving region using optimization algorithm solution present image and background image, can obtain
Complete target area is obtained, needs of the intelligent monitor system for foreground moving extracted region precision is fully met, proposes to be based on
Optimize the background subtraction method of solution, present image is compared into background image, obtains frame difference image Dk+1,
Dk+1=| Fk+1-Bk+1| (19)
Wherein, Fk+1For present image, Bk+1For background image,
Five, the real-time update of background image, by formula (20) to result images Dk+1Binary conversion treatment is carried out, obtains prospect
Using given threshold T, single human body attitude target is extracted using binary image for moving region,
Wherein, T is given threshold, Mk+1It is binary image, if Dk+1Represent there is no moving target in scene during≤T,
Dk+1Represent there is humanbody moving object appearance in scene during ＞ T, in background image Bk+1Under conditions of stabilization, by by present image
Single human body attitude target can accurately be extracted by carrying out difference operation with background image,
Intelligent monitor system is working long hours, and system must in a certain time interval carry out more background
Newly, the quality of background image can be improved by context update, formula (21) realizes background area update, is 0 in binary image
When just carry out the real-time update of background image, update without background when being 1,
Wherein, ρ ∈ [0,1] be weighting coefficient, updated background image can react current background state change and
Renewal rate.
The trans-regional single human body attitude target detection extracting method of a kind of video provided above the embodiment of the present invention, into
It has gone and has been discussed in detail, specific case used herein is expounded the principle of the present invention and embodiment, implements above
The explanation of example is merely used to help understand the method and its core concept of the present invention；Meanwhile for the general technology people of this field
Member, thought according to the present invention, there will be changes in specific embodiments and applications, in conclusion this explanation
Book content should not be construed as limiting the invention.
Claims (1)
1. the trans-regional single human body attitude target detection extracting method of video, it is characterised in that：Single body attitude motion target image
Extraction process includes the following steps：
One, the multi-frame video image of the trans-regional moving target of multiple-camera human body sport video is acquired from intelligent monitor system,
Establish moving object detection Optimized model：
Wherein, X ∈ Rmn×KFor video sequence；B and F is the background and prospect of video, and prospect is the moving target to be looked for；||·|
|·For nuclear norm, the norm calculation be singular values of a matrix and, the effect that nuclear norm is selected in formula (1) is constraint background matrix
Order, | | | |1For 1 norm,
Two, compressed sensing processing is carried out to multi-frame video image, video sequence is compressed while video sequence acquires,
Video foreground and background reconstruction are done directly from compression observing matrix A, and with robustness, can pass through to realize
The three-dimensional circular method of sampling obtains compression observing matrix A, by compression sampling expression formulaIt is divided into two sampling steps
Suddenly:Random convolution process Rk=Cxk；Random sampling procedure SkPiRk=ak, (k=1,2 .., K) enables W=0, above-mentioned two is adopted
Sample step is dissolved into optimization object function, and moving object detection Optimized model is as follows：
R=CX, SkPiRk=ak(k=1,2 ..K) (2)
Wherein, αiFor weighted value (i=1.2,3), DiFor difference operator, W is motion artifacts, and C is circular matrix, xkFor video frame,
R is cycle observing matrix, the initial value of matrix W is set as 0 during (2) formula of solution, when the optimization for solving X, B and F is estimated
Motion artifacts matrix W is calculated after evaluation again, the optimization problem of X, B and F are then solved, when solving-optimizing problem obtains X, B and F
Motion artifacts W can be distinguished, this ensure that the robustness of B and F segmentations,
In order to solve formula (2), it is more to newly arrive to realize while weigh by the alternating iteration of video sequence, background and prospect to carry algorithm
Structure, first image reconstruction procedure are in the reconstruction accuracy for constantly updating video sequence X；Then it is to pass through to carry out image segmentation process
Each video is more newly arrived the segmentation of realization prospect and background, in image reconstruction procedure, by solving following moving object detections
Optimized model realizes the reconstruct of video sequence,
Optimization problem involved by formula (3) can be solved by augmented vector approach, in solution procedure, need first structure
The Lagrange's equation of formula (3) is built,
SkPiRk=ak(k=1,2 ..., K) (4)
In Lagrange's equation is built, μi=(i=1,2,3) and ρ is Lagrange multiplier matrix, passes through Lagrange's equation
Structure, by the required majorization of solutions problem of formula (3), pass through the Lagrangian and Lagrange in iteration newer (4)
Multiplier is solved, and Lagrangian and Lagrange multiplier renewal process are
Utilize D in alternately newer (5)iX (i=1.2.3) is video sequence matrix of variables, and R and X are used alternately newer
Specific implementation process is:
Formula (6) is solved by the update of (9) formula：
Wherein:Function Sa() is threshold value iteration operator, and for scalar x, which is specifically defined as
Sa(x)=sign (x) max x | and-a, 0 } (10)
Wherein:A is iteration threshold, and when processing array threshold value iteration updates, we carry out threshold value to matrix using formula (10) and change
Generation update, after matrix update is completed, output result updates matrix for threshold value, and formula (7) solution can be completed by following step:
Wherein:Rk=(k=1,2 ... it is n) the kth row in cycle observing matrix R, PiCSaDisplacement square is observed for stochastical sampling
Battle array, for representing cycle observing matrix RkIn which element needs be iterated update, PiCSaIt is by stochastical sampling matrix Si
And generate, the middle X updates of formula (8) are realized by solving least square problem,
During display foreground background segment, image reconstruction update result X fixed firstk+1, then by solving following formula (13)
Optimization problem to reconstruct prospect and background simultaneously,
It is by the Lagrangian that formula (13) is established
Wherein:For H Lagrange multiplier matrixes, ＜, ＞ matrix inner products, equally using augmentation Lagrangian Arithmetic to formula
(13) it is solved, process is as follows:
HK+1=HK+λ(XK+1-BK+1-FK+1) (16)
Formula (15) equally solves it using alternately update, display foreground context update process is as follows,
Three, it is more newly arrived by the alternating iteration of video sequence, background and prospect and realizes while reconstruct, formula is solved using restructing algorithm
(2) detailed process is as follows:
Input:C,Pi,Sk,A
Output:X,B,F
Initial value is set as:
B0=F0=R=H0=ρ0=0
When meeting condition:||Xk+1-Xk||F/||Xk||F＞ 10-6It continues cycling through
Image reconstruction procedure:Update:DiX,R
Update:X
Display foreground background segment process is as follows:
Bk+1=Di(Xk+1-Fk-Bk+Hk)
Fk+1=Sk(Xk+1-Bk+1-Fk+Hk)
K=k+1
End loop,
Four, background image is compared using the present image picture of target detection optimization solution, obtains frame difference image：
Present image is compared into background image, obtains frame difference image Dk+1,
Dk+1=| Fk+1-Bk+1| (19)
Wherein, Fk+1For present image, Bk+1For background image,
Five, the real-time update of background image, by formula (20) to result images Dk+1Binary conversion treatment is carried out, obtains foreground moving
Using given threshold T, single human body attitude target is extracted using binary image for region,
Wherein, T is given threshold, Mk+1It is binary image, if Dk+1Represent there is no moving target, D in scene during≤Tk+1＞
Represent there is humanbody moving object appearance in scene during T, in background image Bk+1Under conditions of stabilization, by by present image with the back of the body
Scape image, which carries out difference operation, can accurately extract single human body attitude target,
Formula (21) realizes background area update, the real-time update of background image is just carried out when binary image is 0, when being 1
Without the update of background,
Wherein, ρ ∈ [0,1] are weighting coefficient, and updated background image can react current background state change and update
Rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711255086.2A CN108133487A (en) | 2017-12-04 | 2017-12-04 | The trans-regional single human body attitude target detection extracting method of video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711255086.2A CN108133487A (en) | 2017-12-04 | 2017-12-04 | The trans-regional single human body attitude target detection extracting method of video |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108133487A true CN108133487A (en) | 2018-06-08 |
Family
ID=62390086
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711255086.2A Pending CN108133487A (en) | 2017-12-04 | 2017-12-04 | The trans-regional single human body attitude target detection extracting method of video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108133487A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109033963A (en) * | 2018-06-22 | 2018-12-18 | 王连圭 | The trans-regional human motion posture target identification method of multiple-camera video |
| CN109543650A (en) * | 2018-12-04 | 2019-03-29 | 钟祥博谦信息科技有限公司 | Warehouse intelligent control method and system |
| CN110287917A (en) * | 2019-06-28 | 2019-09-27 | 广东电网有限责任公司 | The security management and control system and method in capital construction building site |
| CN110703145A (en) * | 2019-09-20 | 2020-01-17 | 广东石油化工学院 | Transformer vibration sound signal reconstruction method and system by using multiple optimization theories |
| CN112529784A (en) * | 2019-09-18 | 2021-03-19 | 华为技术有限公司 | Image distortion correction method and device |
| CN112616023A (en) * | 2020-12-22 | 2021-04-06 | 荆门汇易佳信息科技有限公司 | Multi-camera video target tracking method in complex environment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110222781A1 (en) * | 2010-03-15 | 2011-09-15 | U.S. Government As Represented By The Secretary Of The Army | Method and system for image registration and change detection |
| CN104243916A (en) * | 2014-09-02 | 2014-12-24 | 江苏大学 | Moving object detecting and tracking method based on compressive sensing |
| CN106709436A (en) * | 2016-12-08 | 2017-05-24 | 华中师范大学 | Cross-camera suspicious pedestrian target tracking system for rail transit panoramic monitoring |
-
2017
- 2017-12-04 CN CN201711255086.2A patent/CN108133487A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110222781A1 (en) * | 2010-03-15 | 2011-09-15 | U.S. Government As Represented By The Secretary Of The Army | Method and system for image registration and change detection |
| CN104243916A (en) * | 2014-09-02 | 2014-12-24 | 江苏大学 | Moving object detecting and tracking method based on compressive sensing |
| CN106709436A (en) * | 2016-12-08 | 2017-05-24 | 华中师范大学 | Cross-camera suspicious pedestrian target tracking system for rail transit panoramic monitoring |
Non-Patent Citations (2)
| Title |
|---|
| 吴君钦 等: ""图像融合的运动目标检测算法研究"", 《计算机工程与设计》 * |
| 康彬: ""面向智能监控的目标检测、追踪及图像融合算法研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109033963A (en) * | 2018-06-22 | 2018-12-18 | 王连圭 | The trans-regional human motion posture target identification method of multiple-camera video |
| CN109033963B (en) * | 2018-06-22 | 2021-07-06 | 王连圭 | Multi-camera video cross-region human motion posture target recognition method |
| CN109543650A (en) * | 2018-12-04 | 2019-03-29 | 钟祥博谦信息科技有限公司 | Warehouse intelligent control method and system |
| CN110287917A (en) * | 2019-06-28 | 2019-09-27 | 广东电网有限责任公司 | The security management and control system and method in capital construction building site |
| CN110287917B (en) * | 2019-06-28 | 2024-04-05 | 广东电网有限责任公司 | Safety control system and method for construction site |
| CN112529784A (en) * | 2019-09-18 | 2021-03-19 | 华为技术有限公司 | Image distortion correction method and device |
| CN110703145A (en) * | 2019-09-20 | 2020-01-17 | 广东石油化工学院 | Transformer vibration sound signal reconstruction method and system by using multiple optimization theories |
| CN110703145B (en) * | 2019-09-20 | 2021-09-21 | 广东石油化工学院 | Transformer vibration sound signal reconstruction method and system by using multiple optimization theories |
| CN112616023A (en) * | 2020-12-22 | 2021-04-06 | 荆门汇易佳信息科技有限公司 | Multi-camera video target tracking method in complex environment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108133487A (en) | The trans-regional single human body attitude target detection extracting method of video | |
| Dabral et al. | Learning 3d human pose from structure and motion | |
| Poggi et al. | Towards real-time unsupervised monocular depth estimation on cpu | |
| CN107481270B (en) | Table tennis target tracking and trajectory prediction method, device, storage medium and computer equipment | |
| CN107358623B (en) | Relevant filtering tracking method based on significance detection and robustness scale estimation | |
| CN104167016B (en) | A kind of three-dimensional motion method for reconstructing based on RGB color and depth image | |
| CN108921062B (en) | Gait recognition method combining multiple gait features and cooperative dictionary | |
| Liu et al. | Using unsupervised deep learning technique for monocular visual odometry | |
| CN104821010A (en) | Binocular-vision-based real-time extraction method and system for three-dimensional hand information | |
| CN104794737A (en) | Depth-information-aided particle filter tracking method | |
| Zheng et al. | Deep learning for event-based vision: A comprehensive survey and benchmarks | |
| CN106846372B (en) | Human motion quality visual analysis and evaluation system and method thereof | |
| CN109035307B (en) | Set area target tracking method and system based on natural light binocular vision | |
| CN109033963A (en) | The trans-regional human motion posture target identification method of multiple-camera video | |
| 杜绪伟 et al. | Real-time hand tracking based on YOLOv4 model and Kalman filter | |
| Napolean et al. | Heart rate estimation in intense exercise videos | |
| Li | Research on camera-based human body tracking using improved cam-shift algorithm | |
| Jaeggli et al. | Multi-activity tracking in lle body pose space | |
| CN113158942A (en) | Segmentation algorithm and device for detecting motion human behavior | |
| CN112099330B (en) | Holographic human body reconstruction method based on external camera and wearable display control equipment | |
| CN104182990B (en) | A kind of Realtime sequence images motion target area acquisition methods | |
| Li et al. | Adaptive object tracking algorithm based on eigenbasis space and compressive sampling | |
| Abdelali et al. | A new moving object tracking method using particle filter and probability product kernel | |
| Jin et al. | Research on Human Action Recognition Based on Global-Local Features of Video | |
| Wang et al. | RBF-Based 3D Visual Detection Method for Chinese Martial Art Wrong Movements |
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 | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wang Liangui Inventor before: Wang Liangui Inventor before: Wang Yancheng Inventor before: Wang Benyan Inventor before: Wang Wenbin Inventor before: Wang Lian |
|
| CB03 | Change of inventor or designer information | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180608 |
|
| RJ01 | Rejection of invention patent application after publication |