CN104933734B - A kind of human body attitude data fusion method based on more kinect - Google Patents
A kind of human body attitude data fusion method based on more kinect Download PDFInfo
- Publication number
- CN104933734B CN104933734B CN201510363869.7A CN201510363869A CN104933734B CN 104933734 B CN104933734 B CN 104933734B CN 201510363869 A CN201510363869 A CN 201510363869A CN 104933734 B CN104933734 B CN 104933734B
- Authority
- CN
- China
- Prior art keywords
- kinect
- data
- skeleton
- skeleton point
- mrow
- 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.)
- Expired - Fee Related
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
- G06T7/285—Analysis of motion using a sequence of stereo image pairs
-
- 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
- G06T7/251—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
- G06T2207/10021—Stereoscopic video; Stereoscopic image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of human body attitude data fusion method based on more kinect, by building the data collecting system containing more kinect, human body attitude bone information is acquired, data fusion and data prediction are carried out respectively for the classification of gathered information, so as to obtain complete human body attitude bone information, avoiding kinect and being interfered causes skeleton data saltus step occur, or personage causes the detection incomplete problem of skeleton data such as posture missing from blocking, and is that gesture recognition, man-machine interaction, virtual reality etc. are provided convenience in post-processing.
Description
Technical field
The invention belongs to human body attitude detection and gesture recognition technical field, it is related to a kind of human body appearance based on more kinect
State data fusion method.
Background technology
With the development in epoch, the mode of man-machine interaction is also more and more, can not only be exchanged by language,
Body language is also indispensable part, and a branch of computer vision field is exactly study human body behavior posture specific
Implication.However, it is understood that human body behavior posture, first has to get the specific posture of personage, conventional mode is using common
Camera, then use pattern identification carries out analysis identification to the character contour in video, but this mode accuracy rate can't
Meets the needs of man-machine interaction.
Under this application demand, Kinect arises at the historic moment, it be game host Xbox360 of the Microsoft under it and
A kind of motion perception input equipment that windows platform provides.Kinect is really one and utilizes three-dimensional fix technology
Body-sensing camera, the functions such as real-time action seizure, image identification, phonetic entry, speech recognition and community interactive can be carried out.Cause
This user can get personage's framework information using Kinect with the SDK of Microsoft, and each skeleton point is a three-dimensional coordinate
Information.But Kinect using when need user plane to Kinect, the whole human body attitude of personage could be obtained, so Kinect
The problem of for blocking certainly, does not have good solution.
The content of the invention
It is an object of the invention to provide a kind of human body attitude data fusion method based on more kinect, solve existing
When extracting skeleton information, there is saltus step or occurred from posture missing when blocking kinect in the skeleton data that is interfered
Problem.
The technical scheme is that a kind of human body attitude data fusion method based on more kinect, specific steps are such as
Under:
Step 1, data collecting system is built:
It is orthogonal to put two kinect, make it towards shooting area, kinect1 and kinect 2 respectively with computer a and electricity
Brain b connections, LAN is established between computer a and computer b connections, to carry out data transmission;
Step 2, data acquisition:
The skeleton data that kinect 2 is obtained is sent to computer a by open system, computer b by LAN in real time, simultaneously
Computer a obtains kinect 1 skeleton data in real time, wherein each skeleton data includes G(i,j)And F(i,j)Two parts:G(i,j)It is
J-th of bone site coordinate of human body using i-th of kinect as the origin of coordinates;F(i,j)It is whether i-th of kinect traces into people
The label information of j-th of skeleton point of body, F(i,j)Be worth for 0 when, represent i-th of kinect do not trace into j-th of bone site;
F(i,j)Be worth for 1 when, represent i-th of kinect trace into j-th of bone site;
Wherein i represents kinect numbering, and i is 1 or 2;The numbering of j expression skeleton points, 0<J≤bone points;
Step 3, data fusion mode is selected:
Classified for kinect 1 and kinect 2 skeleton datas obtained, determine different pieces of information convergence strategy;
Step 4, data fusion:
Data fusion is carried out for the skeleton data for meeting data fusion condition;
Step 5, data prediction:
Skeleton data for not meeting data fusion condition, ask simple average using the movement change amount to coordinate data
Value is predicted.
The features of the present invention also resides in,
Different Strategy of data fusion described in step 3, it is specially:
1) for meet A=j | F(1,j)=1 } skeleton point, it is believed that obtained the skeleton point information of the point, ignored
The skeleton point that kinect 2 is detected;
2) for meet B=j | F(1,j)=0, F(2,j)=1 } skeleton point, into step 4, data fusion is carried out;
3) for meet C=j | F(1,j)=0, F(2,j)=0 } skeleton point, into step 5, data prediction is carried out.
Step 4 is specially:
4.1 find the skeleton point minimum subscript that two kinect are traced into, that is, ask and meet F(1,j)=1 and F(2,j)=1
Skeleton point, using wherein minimum j values as the datum mark of fused data, it is designated as skeleton point t:
T=min j | F(1,j)=1, F(2,j)=1,0 < j≤skeleton point number }
If not finding the t of the condition of satisfaction, step 5 is directly entered, to all F(1,j)It is pre- that=0 skeleton point carries out data
Survey;
4.2 for all F(2,j)=1 skeleton point, skeleton point that all kinect 2 are traced into is calculated relative to skeleton point
T coordinate offset amount εj:
εj=G(2,j)-G(2,t),j∈B
Wherein G(2,j)For kinect 2 skeleton point j position coordinates, B is the bone for meeting data fusion condition in step 3
Bone point j set, G(2,t)For kinect 2 skeleton point t position coordinates;
4.3 according to coordinate offset amount εjThe undetected bone site data G of all kinect 1 are obtained in calculating(1,j), and
It is 1 by the mark position of these skeleton points, i.e., by F(1,j)It is set to 1:
G(1,j)=G(1,t)+εj,j∈B
Wherein G(1,j)For kinect 1 skeleton point j position coordinates, B is the bone for meeting data fusion condition in step 3
Bone point j set, G(1,t)For kinect 1 skeleton point t position coordinates.
Being used in step 5 asks simple average value to be predicted the movement change amount of coordinate data, comprises the following steps that:
5.1 seek the average value of the change in displacement of the bone site of continuous n frames:
Wherein, T is current frame number;K is frame shifting amount;N is the frame number for needing calculation position to change;δ(j,T)Represent jth
Average value of the individual skeleton point in the preceding n frames change in displacement of T frames;G(1,j,T-k)Represent kinect1 j-th of skeleton point in T-k
The position coordinates of frame, G(1,j,T-k-1)Represent kinect1 position coordinates of j-th of skeleton point in T-k-1 frames;
5.2 try to achieve bone site at the time of needing prediction:
G(1,j,T)=G(1,j,T-1)+δ(j,T)
Wherein, G(1,j,T)J-th of the skeleton point position coordinates of T frames for needing to predict for kinect1, G(1,j,T-1)For
Kinect1 needs j-th of the skeleton point position coordinates of T-1 frames predicted.
The invention has the advantages that skeleton information is acquired by using more Kinect, and for adopting
The information category collected, data fusion sum is carried out it was predicted that being extracted complete human body attitude bone information, is solved existing
Kinect be interfered skeleton data occur saltus step or occur from block when posture missing the problem of, be post-processing in posture
Identification, man-machine interaction, virtual reality etc. are provided convenience.
Brief description of the drawings
Fig. 1 is a kind of hardware chart of the human body attitude data fusion method based on more kinect of the present invention.
In figure, the shooting area of 1.kinect 1,2.kinect 2,3., 4. computer a, 5. computer b.
Embodiment
The present invention is described in detail with reference to the accompanying drawings and detailed description.
A kind of human body attitude data fusion method based on more kinect of the present invention, specifically implements according to the following steps:
Step 1, data collecting system is built:
As shown in figure 1, orthogonal put two kinect, making it, shooting area 3 is by two kinect towards shooting area 3
Camera watch region determine that kinect1 1 and kinect 22 be connected with computer a4 and computer b5 respectively, computer a4 and computer b5 companies
LAN is established between connecing, to carry out data transmission.
Acquiescence computer a is primary processor, during system operation, personage's appearance that computer b in real time can detect Kinect 2
State skeleton data is sent to computer a by LAN, while computer a can also obtain kinect 1 posture skeleton data in real time.
Then computer a can call the more kinect of blending algorithm progress data fusion simultaneously.
Step 2, data acquisition:
Open system, 20 skeleton datas that computer b in real time obtains kinect 2 are sent to computer a by LAN,
Computer a obtains kinect 1 skeleton data in real time simultaneously.Wherein each skeleton data includes G(i,j)And F(i,j)Two parts:
G(i,j)It is j-th of bone site coordinate of human body using i-th of kinect as the origin of coordinates;F(i,j)Be i-th kinect whether with
Label information of the track to j-th of skeleton point of human body.F(i,j)Be worth for 0 when, represent i-th of kinect do not trace into j-th of bone
Position;F(i,j)Be worth for 1 when, represent i-th of kinect trace into j-th of bone site.
Wherein i represents kinect numbering, and i is 1 or 2;The numbering of j expression skeleton points, 0<J≤bone points.
Step 3, data fusion mode is selected:
Because kinect 1 and kinect 2 coordinate information are all and the data fusions using respective kinect as the origin of coordinates
Purpose be merge kinect 1 and kinect 2 skeleton data, to complete a complete human body attitude data, therefore adopt
Convergence strategy is as follows:
1) for meet A=j | F(1,j)=1 } skeleton point, it is believed that obtained the skeleton point information of the point, will ignore
The skeleton point that kinect 2 is detected;
2) for meet B=j | F(1,j)=0, F(2,j)=1 } skeleton point, into step 4, data fusion is carried out;
3) for meet C=j | F(1,j)=0, F(2,j)=0 } skeleton point, into step 5, data prediction is carried out.
Step 4, data fusion:
4.1 find the skeleton point minimum subscript that two kinect are traced into, that is, ask and meet F(1,j)=1 and F(2,j)=1
Skeleton point, using wherein minimum j values as the datum mark of fused data, it is designated as skeleton point t.Benchmark skeleton point t's asks for formula such as
Under:
T=min j | F(1,j)=1, F(2,j)=1,0 < j≤skeleton point number } (1)
If not finding the t of the condition of satisfaction, step 5 is directly entered, to all F(1,j)It is pre- that=0 skeleton point carries out data
Survey;
4.2 for all F(2,j)=1 skeleton point, skeleton point that all kinect 2 are traced into is calculated relative to benchmark bone
Bone point t coordinate offset amount εj:
εj=G(2,j)-G(2,t),j∈B (2)
Wherein G(2,j)For kinect 2 skeleton point j position coordinates, B is the bone for meeting data fusion condition in step 3
Bone point j set, G(2,t)For kinect 2 skeleton point t position coordinates;
4.3 according to coordinate offset amount εjThe undetected bone site data G of all kinect 1 are obtained in calculating(1,j), and
It is 1 by the mark position of these skeleton points, i.e., by F(1,j)It is set to 1:
G(1,j)=G(1,t)+εj,j∈B (3)
Wherein G(1,j)For kinect 1 skeleton point j position coordinates, B is the bone for meeting data fusion condition in step 3
Bone point j set, G(1,t)For kinect 1 skeleton point t position coordinates.
Step 5, data prediction:
After step 3 fusing stage, there may be bony segment coordinate information not get, these bones are sat
Mark information is predicted, i.e., to all F(1,j)Still it is predicted for 0 skeleton point coordinate information.Because consider it is ageing and
The continuity of figure action, the present invention ask simple average value to be predicted using the movement change amount to coordinate data, specific step
It is rapid as follows:
5.1 seek the average value of the change in displacement of the bone site of continuous n frames:
Wherein, T is current frame number;K is frame shifting amount;N is the frame number for needing calculation position to change;δ(j,T)Represent jth
Average value of the individual skeleton point in the preceding n frames change in displacement of T frames;G(1,j,T-k)Represent kinect1 j-th of skeleton point in T-k
The position coordinates of frame, G(1,j,T-k-1)Represent kinect1 position coordinates of j-th of skeleton point in T-k-1 frames.
5.2 try to achieve bone site at the time of needing prediction:
G(1,j,T)=G(1,j,T-1)+δ(j,T) (5)
Wherein, G(1,j,T)J-th of the skeleton point position coordinates of T frames for needing to predict for kinect1, G(1,j,T-1)For
Kinect1 needs j-th of the skeleton point position coordinates of T-1 frames predicted.
By above step, the position coordinates of all bones is all tracked or predicts, and has obtained complete human body attitude
Bone information, solve and do not detect asking for personage's bone information when blocking certainly occurs in personage and Kinect is interfered
Topic, it is that gesture recognition, man-machine interaction, virtual reality etc. are provided convenience in post-processing.
Claims (2)
1. a kind of human body attitude data fusion method based on more kinect, it is characterised in that comprise the following steps that:
Step 1, data collecting system is built:
It is orthogonal to put two kinect, it is connected respectively with computer a and computer b towards shooting area, kinect1 and kinect 2
Connect, LAN is established between computer a and computer b connections, to carry out data transmission;
Step 2, data acquisition:
The skeleton data that kinect 2 is obtained is sent to computer a, while computer a by open system, computer b by LAN in real time
Kinect 1 skeleton data is obtained in real time, wherein each skeleton data includes G(i,j)And F(i,j)Two parts:G(i,j)It is with i-th
Individual kinect is j-th of bone site coordinate of human body of the origin of coordinates;F(i,j)It is whether i-th of kinect traces into human body jth
The label information of individual skeleton point, F(i,j)Be worth for 0 when, represent i-th of kinect do not trace into j-th of bone site;F(i,j)Value
For 1 when, represent i-th of kinect trace into j-th of bone site;
Wherein i represents kinect numbering, and i is 1 or 2;The numbering of j expression skeleton points, 0<J≤bone points;
Step 3, data fusion mode is selected:
Classified for kinect 1 and kinect 2 skeleton datas obtained, determine different pieces of information convergence strategy, be specially:
3.1 for meet A=j | F(1,j)=1 } skeleton point, it is believed that obtained the skeleton point information of the point, ignored kinect 2
The skeleton point detected;
3.2 for meet B=j | F(1,j)=0, F(2,j)=1 } skeleton point, into step 4, data fusion is carried out;
3.3 for meet C=j | F(1,j)=0, F(2,j)=0 } skeleton point, into step 5, data prediction is carried out;
Step 4, data fusion:
Data fusion is carried out for the skeleton data for meeting data fusion condition;
4.1 find the skeleton point minimum subscript that two kinect are traced into, that is, ask and meet F(1,j)=1 and F(2,j)=1 bone
Point, using wherein minimum j values as the datum mark of fused data, it is designated as skeleton point t:
T=min j | F(1,j)=1, F(2,j)=1,0 < j≤skeleton point number }
If not finding the t of the condition of satisfaction, step 5 is directly entered, to all F(1,j)=0 skeleton point carries out data prediction;
4.2 for all F(2,j)=1 skeleton point, skeleton point that all kinect 2 are traced into is calculated relative to skeleton point t's
Coordinate offset amount εj:
εj=G(2,j)-G(2,t),j∈B
Wherein G(2,j)For kinect 2 skeleton point j position coordinates, B is the skeleton point j for meeting data fusion condition in step 3
Set, G(2,t)For kinect 2 skeleton point t position coordinates;
4.3 according to coordinate offset amount εjThe undetected bone site data G of all kinect 1 are obtained in calculating(1,j), and by this
The mark position of a little skeleton points is 1, i.e., by F(1,j)It is set to 1:
G(1,j)=G(1,t)+εj,j∈B
Wherein G(1,j)For kinect 1 skeleton point j position coordinates, B is the skeleton point j for meeting data fusion condition in step 3
Set, G(1,t)For kinect 1 skeleton point t position coordinates;
Step 5, data prediction:
Skeleton data for not meeting data fusion condition, simple average value is asked to enter using the movement change amount to coordinate data
Row prediction.
A kind of 2. human body attitude data fusion method based on more kinect according to claim 1, it is characterised in that step
Being used described in rapid 5 asks simple average value to be predicted the movement change amount of coordinate data, comprises the following steps that:
5.1 seek the average value of the change in displacement of the bone site of continuous n frames:
<mrow>
<msub>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<mi>j</mi>
<mo>,</mo>
<mi>T</mi>
<mo>)</mo>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mo>(</mo>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</msubsup>
<mo>(</mo>
<msub>
<mi>G</mi>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>,</mo>
<mi>j</mi>
<mo>,</mo>
<mi>T</mi>
<mo>-</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>G</mi>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>,</mo>
<mi>j</mi>
<mo>,</mo>
<mi>T</mi>
<mo>-</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mi>n</mi>
</mfrac>
</mrow>
Wherein, T is current frame number;K is frame shifting amount;N is the frame number for needing calculation position to change;δ(j,T)Represent j-th of bone
Average value of the bone o'clock in the preceding n frames change in displacement of T frames;G(1,j,T-k)Represent kinect1 j-th of skeleton point in T-k frames
Position coordinates, G(1,j,T-k-1)Represent kinect1 position coordinates of j-th of skeleton point in T-k-1 frames;
5.2 try to achieve bone site at the time of needing prediction:
G(1,j,T)=G(1,j,T-1)+δ(j,T)
Wherein, G(1,j,T)J-th of the skeleton point position coordinates of T frames for needing to predict for kinect1, G(1,j,T-1)Needed for kinect1
J-th of skeleton point position coordinates of the T-1 frames to be predicted.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510363869.7A CN104933734B (en) | 2015-06-26 | 2015-06-26 | A kind of human body attitude data fusion method based on more kinect |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510363869.7A CN104933734B (en) | 2015-06-26 | 2015-06-26 | A kind of human body attitude data fusion method based on more kinect |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104933734A CN104933734A (en) | 2015-09-23 |
CN104933734B true CN104933734B (en) | 2017-11-28 |
Family
ID=54120887
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510363869.7A Expired - Fee Related CN104933734B (en) | 2015-06-26 | 2015-06-26 | A kind of human body attitude data fusion method based on more kinect |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104933734B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105447889A (en) * | 2015-11-19 | 2016-03-30 | 北京理工大学 | Remote virtual opera command system based body feeling |
CN105740450A (en) * | 2016-02-03 | 2016-07-06 | 浙江大学 | Multi-Kinect based 3D human body posture database construction method |
CN106372858A (en) * | 2016-09-12 | 2017-02-01 | 成都集致生活科技有限公司 | Recruitment system of building industry, and application method thereof |
CN106981075A (en) * | 2017-05-31 | 2017-07-25 | 江西制造职业技术学院 | The skeleton point parameter acquisition devices of apery motion mimicry and its recognition methods |
CN107274438B (en) * | 2017-06-28 | 2020-01-17 | 山东大学 | Single Kinect multi-person tracking system and method supporting mobile virtual reality application |
CN107577451B (en) * | 2017-08-03 | 2020-06-12 | 中国科学院自动化研究所 | Multi-Kinect human body skeleton coordinate transformation method, processing equipment and readable storage medium |
CN107563295B (en) * | 2017-08-03 | 2020-07-28 | 中国科学院自动化研究所 | Multi-Kinect-based all-dimensional human body tracking method and processing equipment |
CN107993249A (en) * | 2017-08-23 | 2018-05-04 | 北京航空航天大学 | A kind of body gait data fusion method based on more Kinect |
CN110866417A (en) * | 2018-08-27 | 2020-03-06 | 阿里巴巴集团控股有限公司 | Image processing method and device and electronic equipment |
CN109373993A (en) * | 2018-10-09 | 2019-02-22 | 深圳华侨城文化旅游科技股份有限公司 | A kind of positioning system and method based on more somatosensory devices |
CN109875562A (en) * | 2018-12-21 | 2019-06-14 | 鲁浩成 | A kind of human somatotype monitoring system based on the more visual analysis of somatosensory device |
CN111582081A (en) * | 2020-04-24 | 2020-08-25 | 西安交通大学 | Multi-Kinect serial gait data space-time combination method and measuring device |
CN112200126A (en) * | 2020-10-26 | 2021-01-08 | 上海盛奕数字科技有限公司 | Method for identifying limb shielding gesture based on artificial intelligence running |
CN112891922B (en) * | 2021-03-18 | 2022-11-22 | 山东梦幻视界智能科技有限公司 | Virtual reality somatosensory interaction method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103529944A (en) * | 2013-10-17 | 2014-01-22 | 合肥金诺数码科技股份有限公司 | Human body movement identification method based on Kinect |
CN203673527U (en) * | 2014-01-14 | 2014-06-25 | 河海大学常州校区 | Human body three-dimension scanning hardware platform based on kinects |
-
2015
- 2015-06-26 CN CN201510363869.7A patent/CN104933734B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103529944A (en) * | 2013-10-17 | 2014-01-22 | 合肥金诺数码科技股份有限公司 | Human body movement identification method based on Kinect |
CN203673527U (en) * | 2014-01-14 | 2014-06-25 | 河海大学常州校区 | Human body three-dimension scanning hardware platform based on kinects |
Non-Patent Citations (4)
Title |
---|
Estimating human motion from multiple Kinect Sensors;Stylianos Asteriadis 等;《International Conference on Mirage》;20131231;全文 * |
Evaluation of Skeleton Trackers and Gesture Recognition for Human-robot Interaction;Martin Bünger;《Applied Catalysis A General》;20141231;第475卷(第5期);正文Ⅳ Appendices的E Article的Ⅲ DATA ACQUISITON的A. Skeleton Tracking Systems的第1段和C. Dual Kinects、IV. RESULTS AND DISCUSSION的第2段和B. Dual Kinect的第1段 * |
Multi-Kinect Skeleton Fusion for Physical Rehabilitation Monitoring;Saiyi Li 等;《Engineering in Medicine & Biology Society》;20141231;全文 * |
Tracking People across Multiple Non-Overlapping RGB-D Sensors;Emilio J. Almazan 等;《Computer Vision & Pattern Recognition Workshops》;20131231;第13卷(第4期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN104933734A (en) | 2015-09-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104933734B (en) | A kind of human body attitude data fusion method based on more kinect | |
KR101711736B1 (en) | Feature extraction method for motion recognition in image and motion recognition method using skeleton information | |
CN105930767B (en) | A kind of action identification method based on human skeleton | |
CN103941866B (en) | Three-dimensional gesture recognizing method based on Kinect depth image | |
CN110363867B (en) | Virtual decorating system, method, device and medium | |
CN102184541B (en) | Multi-objective optimized human body motion tracking method | |
JP2021524113A (en) | Image processing methods and equipment, imaging equipment, and storage media | |
CN104616028B (en) | Human body limb gesture actions recognition methods based on space segmentation study | |
EP3628380B1 (en) | Method for controlling virtual objects, computer readable storage medium and electronic device | |
CN110570455A (en) | Whole body three-dimensional posture tracking method for room VR | |
CN105159452B (en) | A kind of control method and system based on human face modeling | |
CN107992858A (en) | A kind of real-time three-dimensional gesture method of estimation based on single RGB frame | |
CN108829233B (en) | Interaction method and device | |
Wu et al. | Incorporating motion analysis technology into modular arrangement of predetermined time standard (MODAPTS) | |
CN116052276A (en) | Human body posture estimation behavior analysis method | |
CN115576426A (en) | Hand interaction method for mixed reality flight simulator | |
Huang et al. | Intelligent yoga coaching system based on posture recognition | |
Zou et al. | Automatic reconstruction of 3D human motion pose from uncalibrated monocular video sequences based on markerless human motion tracking | |
Yin et al. | A systematic review on digital human models in assembly process planning | |
CN112602100B (en) | Action analysis device, action analysis method, storage medium, and action analysis system | |
Alexanderson et al. | Towards Fully Automated Motion Capture of Signs--Development and Evaluation of a Key Word Signing Avatar | |
CN112541870A (en) | Video processing method and device, readable storage medium and electronic equipment | |
CN203630717U (en) | Interaction system based on a plurality of light inertial navigation sensing input devices | |
CN111310655A (en) | Human body action recognition method and system based on key frame and combined attention model | |
US11887257B2 (en) | Method and apparatus for virtual training based on tangible interaction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171128 Termination date: 20200626 |
|
CF01 | Termination of patent right due to non-payment of annual fee |