CN104715493B - A kind of method of movement human Attitude estimation - Google Patents

A kind of method of movement human Attitude estimation Download PDF

Info

Publication number
CN104715493B
CN104715493B CN201510128533.2A CN201510128533A CN104715493B CN 104715493 B CN104715493 B CN 104715493B CN 201510128533 A CN201510128533 A CN 201510128533A CN 104715493 B CN104715493 B CN 104715493B
Authority
CN
China
Prior art keywords
mtd
mrow
msub
point
human body
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
Application number
CN201510128533.2A
Other languages
Chinese (zh)
Other versions
CN104715493A (en
Inventor
孔德慧
刘洪林
王少帆
尹宝才
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.)
Beijing University of Technology
Original Assignee
Beijing 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 Beijing University of Technology filed Critical Beijing University of Technology
Priority to CN201510128533.2A priority Critical patent/CN104715493B/en
Publication of CN104715493A publication Critical patent/CN104715493A/en
Application granted granted Critical
Publication of CN104715493B publication Critical patent/CN104715493B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Image Analysis (AREA)

Abstract

The present invention discloses a kind of method of movement human Attitude estimation, can position human skeleton point exactly, effectively obtains the feature that expressiveness is had more in three-dimensional motion human body, and structure is more effective, simple.It includes：(1) pretreatment operation is carried out to depth image data using the method for medium filtering, position demarcation is carried out to human body pixel using based on the Dijkstra's algorithm of geodesic distance；(2) the provincial characteristics point extraction algorithm based on K means clustering algorithms, it is determined that the cluster number in each class is 3,32 posture features are extracted to characterize different human body attitudes；(3) skeletal point position markup information is obtained by PoserPro2012 softwares in the training stage, synthesize the posture feature of 300 frame visual humans and be labelled with standard skeletal point, pass through the posture feature point and standard skeletal point of training sample, the linear regression model (LRM) of posture feature and skeletal point is calculated, to obtain the mapping relations between posture feature and standard skeletal point.

Description

A kind of method of movement human Attitude estimation
Technical field
The invention belongs to computer vision and the technical field of pattern-recognition, estimates more particularly to a kind of movement human posture The method of meter.
Background technology
Human body attitude estimation is an important research direction of computer vision field.Nearly ten years, automatic identification figure As the human body attitude problem in video sequence is always the study hotspot of computer vision field.Human body attitude estimation is promoted to turn into The main reason for research emphasis or electronic equipment develop rapidly and the huge applications market caused by it.Effective place Manage and understand the physical activity in data, it will bring profound influence for the development of human society.Movement human behavioural analysis Purpose is to describe, identify and understand the interbehavior between human action, interpersonal and human and environment, and it is in intelligence Video monitoring, virtual reality, safety, advanced man-machine interaction and image storage and retrieval based on content etc. have extensive Application background.
The posture of human body represents that dimension has two kinds of two and three dimensions.The human body attitude of two dimension refers to human synovial in image two A kind of description of dimensional plane distribution, the human body attitude of traditional two dimensional image are estimated by ambient environmental factors (clothes color, light According to) and the influence blocked it is bigger, while lack the spatial positional information of pixel.Estimate at present towards the human body attitude of two dimensional image In meter method, overwhelming dominance is accounted for based on graph structure model and its improved method.Graph structure model be with graph model structure come Represent the connection between part.Graph structure model by human body be divided into multiple rigid body parts (head, trunk, a pair of upper arm, a pair Underarm, a pair of thighs, a pair of shanks etc.), each position carries out normal indication with a rectangle frame；By artis between two parts It is connected (such as Fig. 1 a).The rectangle frame of corresponding human body can be expressed as vector
L=(x, y, r, s, w) (1)
Wherein (x, y) represents rectangular centre position, and r represents that rectangle represents square relative to the offset angle of vertical direction, s The length of shape frame, w represent rectangle width of frame.The tree-like graph model of human body (such as Fig. 1 b) can be expressed as a non-directed graph
G=(V, E) (2)
Wherein E is the set on all sides in figure, vertex set V={ v1,v2,v3,...,vnIn each element difference It is corresponding human body rigidity position, if two human body viAnd vjIt is connected, then side (v is presenti, vj)∈E.People based on graph model Body Attitude estimation needs the expression of design feature, part to detect and handles the topological structure of complexity.Therefore the side based on graph model Method is not a kind of very efficient estimation method of human posture.For two dimension human body attitude estimation problem, 2014 Alexander Toshev et al. propose the estimation method of human posture based on DNN, and this method is by human body attitude estimation problem Form turns to the regression problem of artis.Obvious lack be present in the human body attitude estimation based on depth convolutional neural networks Fall into：Just for RGB image, depth data is not used, while the network structure that this method uses is complicated (such as Fig. 2) in the extreme, Training effectiveness is low.
At present, 3 D human body Attitude estimation method can generally be divided into two classes.One kind is with Loren Arthur The method that Schwarz et al. is proposed is the method based on geodesic distance and light stream of representative, is that its is right the defects of such algorithm In the mode of the positioning adoption rate method positioning of two level skeletal point (ancon, knee, neck, shoulder, crotch), therefore for difference People's effect of build is less desirable, while the amount of calculation of application light stream is larger it is difficult to meet the higher field of requirement of real-time Close.Another kind of method is the method based on cluster, using Jamie Shotton propose based on the clustering method of random forest as generation Table, each skeletal point is the regressand value of all pixels in such method, and model is complicated and needs substantial amounts of training sample, leads to Cross largely have supervision training process can the ideal element of fixation really for each skeletal point weight.
The content of the invention
The technology of the present invention solves problem：Overcome the deficiencies in the prior art, there is provided a kind of movement human Attitude estimation Method, it can position human skeleton point exactly, effectively obtain the feature that expressiveness is had more in three-dimensional motion human body, structure More effectively, simply.
The present invention technical solution be：The method of this movement human Attitude estimation, comprises the following steps：
(1) pretreatment operation is carried out to depth image data using the method for medium filtering, using based on geodesic distance Dijkstra's algorithm carries out position demarcation to human body pixel；
(2) the provincial characteristics point extraction algorithm based on K- means clustering algorithms, it is determined that the cluster number in each class is 3 It is individual, 32 posture features are extracted to characterize different human body attitudes；
(3) skeletal point position markup information is obtained by the softwares of Poser Pro 2012 in the training stage, 300 frames of synthesis are empty The posture feature of personification is simultaneously labelled with standard skeletal point, passes through the posture feature point and standard skeletal point of training sample, calculates appearance The linear regression model (LRM) of state feature and skeletal point, to obtain the mapping relations between posture feature and standard skeletal point.
The present invention can extremely accurate position human body one by single source point Dijkstra's algorithm based on geodesic distance Level skeletal point (four limbs and head).In order to positioning human body two level skeletal point (neck, elbow, knee joint, hip more easily and effectively Portion) etc., it is proposed that the feature extraction mode based on cluster, the selection mode of this feature by seek a kind of efficient feature and Mapping relations between skeletal point, make this method that there is higher treatment effeciency, can accurately position human body two level skeletal point. This method is started with from the mode of cluster, can effectively obtain the feature that expressiveness is had more in three-dimensional motion human body, structure is more Effectively, simply.
Brief description of the drawings
Fig. 1 a are organizations of human body, and Fig. 1 b are graph structure models.
Fig. 2 is DNN model structure schematic diagrames.
Embodiment
The method of this movement human Attitude estimation, comprises the following steps：
(1) pretreatment operation is carried out to depth image data using the method for medium filtering, using based on geodesic distance Dijkstra's algorithm carries out position demarcation to human body pixel；
(2) the provincial characteristics point extraction algorithm based on K- means clustering algorithms, it is determined that the cluster number in each class is 3 It is individual, 32 posture features are extracted to characterize different human body attitudes；
(3) skeletal point position markup information is obtained by the softwares of Poser Pro 2012 in the training stage, 300 frames of synthesis are empty The posture feature of personification is simultaneously labelled with standard skeletal point, passes through the posture feature point and standard skeletal point of training sample, calculates appearance The linear regression model (LRM) of state feature and skeletal point, to obtain the mapping relations between posture feature and standard skeletal point.
The present invention can extremely accurate position human body one by single source point Dijkstra's algorithm based on geodesic distance Level skeletal point (four limbs and head).In order to positioning human body two level skeletal point (neck, elbow, knee joint, hip more easily and effectively Portion) etc., it is proposed that the feature extraction mode based on cluster, the selection mode of this feature by seek a kind of efficient feature and Mapping relations between skeletal point, make this method that there is higher treatment effeciency, can accurately position human body two level skeletal point. This method is started with from the mode of cluster, can effectively obtain the feature that expressiveness is had more in three-dimensional motion human body, structure is more Effectively, simply.
Preferably, the pretreatment operation of the step (1) includes smooth, denoising, alignment and background rejecting acquisition depth map Human body pixel as in, position demarcation is carried out to human body pixel using based on the Dijkstra's algorithm of geodesic distance：Based on figure The pixel of structure represents, using single source point Dijkstra's algorithm based on geodesic distance, with the geometric center point of human body pixel As source point, calculated, indicate the limb endpoint location of the right-hand man of human body, left and right pin and head；Single source point Di Jiesitela is calculated The construction that initializing adjacency matrix in method has adjacent side meets formula (3)：
(xij, xkt)∈Vt×Vt, | | xij-xkt||2＜ δ ∧ | i-k |≤1 ∧ | j-t |≤1 (3)
Wherein xijRepresent the pixel depth value in (i, j) position, V in imagetIt is human body pixel set, δ represents connection The statistics empirical value of pixel；5 acra points of human body are asked for by single source point Dijkstra's algorithm, using based on geodesic distance Arest neighbors sorting algorithm, using acra point and geometric center as the class heart, realize that human region divides；Pattern class shares 6 classesHuman body acra point position and trunk are represented respectively, whereinFor metastomium, according to formula (4), (9) obtain
WhereinSubscript i representClass, c are representedGeometric center point pixel in class,
||x-y||geodesicRepresent the geodesic distance between pixel.
Preferably, the feature point extraction in the step (2) is the K- means clustering algorithms based on human region：
Posture feature extracts：Using the K- means clustering algorithms based on human body region, in five obtained human body portions PositionIt is interior, use the K- means clustering algorithms that cluster number is three to extract in cluster this five human bodies Heart position feature.
Characteristic dimension：At five acra point positionsIt is interior, use and cluster number as three K- mean clusters Algorithm, obtain 15 human body posture feature points；In order to obtain the global description of human body attitude, added in human body attitude characteristic point The geometric center point of human body pixel is 16 as global description's feature, human body attitude characteristic point, and characteristic point includes its two-dimensional coordinate Value, characteristic dimension 32.
Preferably, a sparse linear projection matrix B is obtained according to formula (5)-(7) in the step (3) so that by X predictions Y error is smaller
Wherein X={ x1..., xnRepresent to cluster the characteristic point sample set of n obtained sample, each
Sample xi={ xi1,xi2...xim, wherein i ∈ { 1,2,3..., n }, m are the Characteristic Numbers 32 of extraction.
Y={ y1..., ynRepresent the n sample that the corresponding skeletal point label of training sample forms, each sample yi= {yi1,yi2...yit, wherein i ∈ { 1,2,3..., n }, t are the numbers of the standard skeleton point coordinates of human body attitude.
In order to ensure the rotational invariance of subspace distance measurement, model adds spin matrix R, and will extract human body bone Frame point problem form turns to rotation sparse regression problem.Preferably, human body bone will be extracted according to formula (8) in the step (3) Frame point form turns to rotation sparse regression problem solving,
The present invention is applied to virtual data derived from the depth image data and Poser Pro 2012 of Kinect2 acquisitions In, and achieve obvious effect.640 × 480 RGBD images are selected in an experiment, and collection environment is interior, gathers light According to for fluorescent lamp.
The at a relatively high degree of accuracy is achieved in generated data test set Poser Pro 2012, tests 100 frame depth maps As data, the root-mean-square error (RMS) and worst error (Max) of generated data are counted, its numerical value is in units of pixel. RMS=10.1305 maximums joint Euclidean distance error is 37.3363, it can be seen that the data sense of reality of synthesis and the degree of accuracy compared with It is good.The present invention compared with the method for Jamie Shotton et al. the random forest proposed from two standards of RMS and MAX all Achieve preferable lifting.
In the actual grade image measurement data that Kinect2 is obtained, this method equally achieves preferable effect.
It is described above, be only presently preferred embodiments of the present invention, any formal limitation not made to the present invention, it is every according to Any simple modification, equivalent change and modification made according to the technical spirit of the present invention to above example, still belong to the present invention The protection domain of technical scheme.

Claims (4)

  1. A kind of 1. method of movement human Attitude estimation, it is characterised in that：Comprise the following steps：
    (1) pretreatment operation is carried out to depth image data using the method for medium filtering, using the Di Jie based on geodesic distance Si Tela algorithms carry out position demarcation to human body pixel；
    (2) the provincial characteristics point extraction algorithm based on K- means clustering algorithms, it is determined that the cluster number in each class is 3, carry 32 dimension posture features are taken to characterize different human body attitudes；
    (3) skeletal point position markup information is obtained by the softwares of Poser Pro 2012 in the training stage, synthesizes 300 frame visual humans Posture feature and be labelled with standard skeletal point, by the posture feature point of training sample and standard skeletal point, it is special to calculate posture Linear regression model (LRM) of the sign point with skeletal point, to obtain the mapping relations between posture feature and standard skeletal point；
    The pretreatment operation of the step (1) includes smooth, denoising, alignment and background and rejects the human body picture obtained in depth image Element, position demarcation is carried out to human body pixel using based on the Dijkstra's algorithm of geodesic distance：Pixel table based on graph structure Show, using single source point Dijkstra's algorithm based on geodesic distance, using the geometric center point of human body pixel as source point, carry out Calculate, indicate the limb endpoint location of the right-hand man of human body, left and right pin and head；Adjoining is initialized in single source point Dijkstra's algorithm The construction that matrix has adjacent side meets formula (3)：
    (xij, xkt)∈Vt×Vt, | | xij-xkt||2＜ δ ∧ | i-k |≤1 ∧ | j-t |≤1 (3)
    Wherein xijRepresent the pixel depth value in (i, j) position, V in imagetIt is human body pixel set, δ represents connected pixel Statistics empirical value；5 acra points of human body are asked for by single source point Dijkstra's algorithm, using based on the nearest of geodesic distance Adjacent sorting algorithm, using acra point and geometric center as the class heart, realize that human region divides；Pixel classification shares 6 classesI= 1,2 ..., 6, human body acra point position and trunk are represented respectively, whereinFor metastomium, obtained according to formula (4), (9)
    <mrow> <msub> <mi>g</mi> <mi>i</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>k</mi> <mi>t</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mi>k</mi> </munder> <mo>|</mo> <mo>|</mo> <msub> <mi>x</mi> <mrow> <mi>k</mi> <mi>t</mi> </mrow> </msub> <mo>-</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>c</mi> </msubsup> <mo>|</mo> <msub> <mo>|</mo> <mrow> <mi>g</mi> <mi>e</mi> <mi>o</mi> <mi>d</mi> <mi>e</mi> <mi>s</mi> <mi>i</mi> <mi>c</mi> </mrow> </msub> <mo>,</mo> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mn>6</mn> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </mrow>
    <mrow> <msub> <mover> <mi>&amp;omega;</mi> <mo>&amp;OverBar;</mo> </mover> <mi>j</mi> </msub> <mo>=</mo> <mo>{</mo> <msub> <mi>x</mi> <mrow> <mi>k</mi> <mi>t</mi> </mrow> </msub> <mo>:</mo> <msub> <mi>g</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>k</mi> <mi>t</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mrow> <mn>1</mn> <mo>&amp;le;</mo> <mi>i</mi> <mo>&amp;le;</mo> <mn>6</mn> </mrow> </munder> <msub> <mi>g</mi> <mi>i</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>k</mi> <mi>t</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>}</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>9</mn> <mo>)</mo> </mrow> </mrow>
    WhereinSubscript i representClass, c are representedGeometric center point pixel in class,
    ||x-y||geodesicRepresent the geodesic distance between pixel.
  2. 2. the method for movement human Attitude estimation according to claim 1, it is characterised in that：Spy in the step (2) Sign point extraction is the K- means clustering algorithms based on human region：
    Posture feature extracts：Using the K- means clustering algorithms based on human body region, in five obtained human bodies In j=1 ..., 5, cluster number is used to calculate cluster centre position to this five human bodies for three K- means clustering algorithms Put,
    Characteristic dimension：At five acra point positionsIn j=1 ..., 5, cluster number is used to be calculated for three K- mean clusters Method, obtain 15 human body posture feature points；In order to obtain the global description of human body attitude, people is added in human body attitude characteristic point The geometric center point of volumetric pixel is 16 as global description's feature, human body attitude characteristic point, and characteristic point includes its two-dimensional coordinate value, Characteristic dimension is 32.
  3. 3. the method for movement human Attitude estimation according to claim 2, it is characterised in that：Basis in the step (3) Formula (5)-(7) obtain a sparse linear projection matrix B, to minimize the prediction for mapping to obtain by projection matrix B by X The error of characteristic point position and fact characteristic point position Y is object function
    <mrow> <mi>X</mi> <mo>=</mo> <mfenced open = "(" close = ")"> <mtable> <mtr> <mtd> <msub> <mi>x</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>12</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>1</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>22</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>2</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>2</mn> </mrow> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mi>m</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow>
    <mrow> <mi>Y</mi> <mo>=</mo> <mfenced open = "(" close = ")"> <mtable> <mtr> <mtd> <msub> <mi>y</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mn>12</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mi>t</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>y</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mn>22</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mi>t</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mo>...</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mn>2</mn> </mrow> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mi>t</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>6</mn> <mo>)</mo> </mrow> </mrow>
    Wherein X={ x1..., xnRepresent to cluster the characteristic point sample set of n obtained sample, each sample xi={ xi1, xi2...xim, wherein i ∈ { 1,2,3..., n }, m=32 be extraction Characteristic Number, Y={ y1..., ynRepresent training sample N sample of corresponding skeletal point label composition, each sample yi={ yi1, yi2...yit, wherein i ∈ 1,2,3..., N }, t is the number of the standard skeleton point coordinates of human body attitude.
  4. 4. the method for movement human Attitude estimation according to claim 3, it is characterised in that：Basis in the step (3) Formula (8) turns to the solution of rotation sparse regression problem by human skeleton point form is extracted,
CN201510128533.2A 2015-03-23 2015-03-23 A kind of method of movement human Attitude estimation Expired - Fee Related CN104715493B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510128533.2A CN104715493B (en) 2015-03-23 2015-03-23 A kind of method of movement human Attitude estimation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510128533.2A CN104715493B (en) 2015-03-23 2015-03-23 A kind of method of movement human Attitude estimation

Publications (2)

Publication Number Publication Date
CN104715493A CN104715493A (en) 2015-06-17
CN104715493B true CN104715493B (en) 2018-01-19

Family

ID=53414787

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510128533.2A Expired - Fee Related CN104715493B (en) 2015-03-23 2015-03-23 A kind of method of movement human Attitude estimation

Country Status (1)

Country Link
CN (1) CN104715493B (en)

Families Citing this family (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105389569B (en) * 2015-11-17 2019-03-26 北京工业大学 A kind of estimation method of human posture
CN106097352A (en) * 2016-06-15 2016-11-09 深圳市易奉亲智慧养老科技有限公司 Human body localization method based on video image and device
CN106485750A (en) * 2016-09-13 2017-03-08 电子科技大学 A kind of estimation method of human posture based on supervision Local Subspace
EP3709271B1 (en) * 2016-09-15 2022-11-02 Google LLC Image depth prediction neural networks
CN106611158A (en) * 2016-11-14 2017-05-03 深圳奥比中光科技有限公司 Method and equipment for obtaining human body 3D characteristic information
CN106778474A (en) * 2016-11-14 2017-05-31 深圳奥比中光科技有限公司 3D human body recognition methods and equipment
CN106909890B (en) * 2017-01-23 2020-02-11 北京工业大学 Human behavior recognition method based on part clustering characteristics
CN106909896B (en) * 2017-02-17 2020-06-30 竹间智能科技（上海）有限公司 Man-machine interaction system based on character personality and interpersonal relationship recognition and working method
CN106955112A (en) * 2017-03-17 2017-07-18 泉州装备制造研究所 Brain wave Emotion recognition method based on Quantum wavelet neural networks model
CN107832713B (en) * 2017-11-13 2021-11-16 南京邮电大学 Human body posture recognition method based on OptiTrack
CN108229332B (en) 2017-12-08 2020-02-14 华为技术有限公司 Bone posture determination method, device and computer readable storage medium
CN108256489B (en) * 2018-01-24 2020-09-25 清华大学 Behavior prediction method and device based on deep reinforcement learning
CN108345869B (en) * 2018-03-09 2022-04-08 南京理工大学 Driver posture recognition method based on depth image and virtual data
CN108549863B (en) * 2018-04-11 2019-11-26 腾讯科技（深圳）有限公司 Human body gesture prediction method, apparatus, equipment and storage medium
CN108527370B (en) * 2018-04-16 2020-06-02 北京卫星环境工程研究所 Human-computer co-fusion safety protection control system based on vision
CN108830150B (en) * 2018-05-07 2019-05-28 山东师范大学 One kind being based on 3 D human body Attitude estimation method and device
CN108830215B (en) * 2018-06-14 2021-07-13 南京理工大学 Dangerous behavior identification method based on personnel skeleton information
CN109684943B (en) * 2018-12-07 2021-03-16 北京首钢自动化信息技术有限公司 Athlete auxiliary training data acquisition method and device and electronic equipment
CN110555412B (en) * 2019-09-05 2023-05-16 深圳龙岗智能视听研究院 End-to-end human body gesture recognition method based on combination of RGB and point cloud
CN111476202B (en) * 2020-04-30 2021-05-25 浙江申汇金融服务外包有限公司 User behavior analysis method and system of security system
CN112232106B (en) * 2020-08-12 2024-07-05 北京工业大学 Two-dimensional to three-dimensional human body posture estimation method
CN113065506B (en) * 2021-04-16 2023-12-26 南京大学 Human body posture recognition method and system
CN113524175B (en) * 2021-06-22 2022-08-12 香港理工大学深圳研究院 Man-machine cooperation method and system based on multi-mode behavior online prediction
CN113706422B (en) * 2021-10-28 2022-03-18 深圳市亚略特科技股份有限公司 Image correction method, device, equipment and medium based on key point acquisition
CN116524088B (en) * 2023-07-03 2023-09-19 深圳星坊科技有限公司 Jewelry virtual try-on method, jewelry virtual try-on device, computer equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102486816A (en) * 2010-12-02 2012-06-06 三星电子株式会社 Device and method for calculating human body shape parameters
CN103150752A (en) * 2013-02-02 2013-06-12 浙江大学 Key mark point-based human body posture sparse reconstruction method
CN104200455A (en) * 2014-06-13 2014-12-10 北京工业大学 Key posture extraction method based on motion statistical feature analysis

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8503720B2 (en) * 2009-05-01 2013-08-06 Microsoft Corporation Human body pose estimation
US9058663B2 (en) * 2012-04-11 2015-06-16 Disney Enterprises, Inc. Modeling human-human interactions for monocular 3D pose estimation

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102486816A (en) * 2010-12-02 2012-06-06 三星电子株式会社 Device and method for calculating human body shape parameters
CN103150752A (en) * 2013-02-02 2013-06-12 浙江大学 Key mark point-based human body posture sparse reconstruction method
CN104200455A (en) * 2014-06-13 2014-12-10 北京工业大学 Key posture extraction method based on motion statistical feature analysis

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Estimating Human 3D Pose from Time-of-Flight Images Based on Geodesic Distances and Optical Flow;Loren Arthur Schwarz et al.;《Automatic Face & Gesture Recognition and Workshops (FG 2011), 2011 IEEE International Conference on》;20110325;第702页A-C节，图2，第703-704页E节 *
基于三维模型的关节式物体姿态估计与跟踪方法研究;崔锦实;《中国优秀博硕士学位论文全文数据库 (博士) 信息科技辑》;20050715;第35页 *

Also Published As

Publication number Publication date
CN104715493A (en) 2015-06-17

Similar Documents

Publication Publication Date Title
CN104715493B (en) A kind of method of movement human Attitude estimation
Yu et al. DS-SLAM: A semantic visual SLAM towards dynamic environments
Alldieck et al. imghum: Implicit generative models of 3d human shape and articulated pose
CN108830150B (en) One kind being based on 3 D human body Attitude estimation method and device
CN108596974B (en) Dynamic scene robot positioning and mapping system and method
Kamal et al. A hybrid feature extraction approach for human detection, tracking and activity recognition using depth sensors
CN107423730B (en) Human gait behavior active detection and recognition system and method based on semantic folding
Gu et al. Multi-person hierarchical 3d pose estimation in natural videos
CN105869178B (en) A kind of complex target dynamic scene non-formaldehyde finishing method based on the convex optimization of Multiscale combination feature
CN108154550A (en) Face real-time three-dimensional method for reconstructing based on RGBD cameras
CN110222665A (en) Human motion recognition method in a kind of monitoring based on deep learning and Attitude estimation
CN106600626B (en) Three-dimensional human motion capture method and system
CN106469465A (en) A kind of three-dimensional facial reconstruction method based on gray scale and depth information
CN103735269B (en) A kind of height measurement method followed the tracks of based on video multi-target
CN103733227A (en) Three-dimensional object modelling fitting &amp; tracking
Zhang et al. Unsupervised skeleton extraction and motion capture from 3D deformable matching
CN113449570A (en) Image processing method and device
CN110310285A (en) A kind of burn surface area calculation method accurately rebuild based on 3 D human body
Ran et al. Applications of a simple characterization of human gait in surveillance
CN106815855A (en) Based on the human body motion tracking method that production and discriminate combine
CN104050685A (en) Moving target detection method based on particle filtering visual attention model
CN102156994B (en) Joint positioning method for single-view unmarked human motion tracking
CN108073855A (en) A kind of recognition methods of human face expression and system
Hu et al. Human interaction recognition using spatial-temporal salient feature
Darujati et al. Facial motion capture with 3D active appearance models

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: 20180119

CF01 Termination of patent right due to non-payment of annual fee