CN107392163A - A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo - Google Patents
A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo Download PDFInfo
- Publication number
- CN107392163A CN107392163A CN201710632589.0A CN201710632589A CN107392163A CN 107392163 A CN107392163 A CN 107392163A CN 201710632589 A CN201710632589 A CN 201710632589A CN 107392163 A CN107392163 A CN 107392163A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- human hand
- identification
- pixel
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/653—Three-dimensional objects by matching three-dimensional models, e.g. conformal mapping of Riemann surfaces
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
- G06F3/0481—Interaction techniques based on graphical user interfaces [GUI] based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance
- G06F3/04815—Interaction with a metaphor-based environment or interaction object displayed as three-dimensional, e.g. changing the user viewpoint with respect to the environment or object
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo proposed in the present invention, its main contents include：Scene modeling, the modeling of identification figure, model evaluation, random optimization, its process is, calculating first to being entered as carrying out human hand three-dimensional color image identification matching figure, then the human hand modeling with skin texture is carried out, the parameter of solving model on the basis of based on particle cluster algorithm, and calculate its region continuity to arrange by evaluation result, the model of highest scoring is used for predicting tracing, and is added in actual scene and is projected.The present invention can handle the hand-type of bending and posture in various degree and model, there is provided the parametric solution method based on particle cluster algorithm, accurately and efficiently human hand and its object interaction are tracked.
Description
Technical field
The present invention relates to limb action analysis field, more particularly, to a kind of human hand based on the imaging of short Baseline Stereo and
Its object interaction tracking.
Background technology
In computer vision field, human hand acts to be recovered to have great importance with the 3-D view of posture.People
No matter be either required for human communication using hand always coming into contacts with physical world, and occupation mode it is various and
It is different.Therefore, a kind of method that is accurate, sane and analyzing and track for hand motion in real time has weight for the mankind
Influence, and have huge business potential or practical value in many fields, such as man-machine interaction, the medical science limbs of Internet of Things
Rehabilitation, sign language motion analysis etc., all urgently developmental research.In addition, in hand-held object posture analysis, the command system of field of safety check
Long-range gesture identification, intelligent robot anthropomorphic emulation and substitute movements design of the mankind in dangerous or uncomfortable working environment
Etc. have very big application prospect.
The tracking of three-dimensional hand gestures is still challenging.Because complex-shaped, details is abundant, posture is changeable, opponent
The modeling of type is not easy to.Simultaneously in terms of details, because the uncertainty of articulated structure and its disunity of length degree
Property, the phenomenon that ambiguous posture often occurs when finding a view and block certainly, and the urgency acted is slow, weight is also to posture
Modeling brings harmful effect, therefore brings difficulty in the train of thought filling process modeled to hand-type.
The present invention proposes a kind of new frame that optimized parameter is solved based on random optimization.It is three-dimensional to human hand color to being entered as
Color image carries out the calculating of identification matching figure, the human hand modeling with skin texture is then carried out, based on particle cluster algorithm
On the basis of solving model parameter, and calculate its region continuity and arranged by evaluation result, the model of highest scoring is used for pre-
Tracking is surveyed, and is added in actual scene and is projected.The present invention can handle the hand-type of bending and posture in various degree and model,
The parametric solution method based on particle cluster algorithm is provided, accurately and efficiently human hand and its object interaction are tracked.
The content of the invention
For solving the problems, such as to carry out human hand and its Object tracking in complex scene, it is an object of the invention to provide one
Human hand and its object interaction tracking of the kind based on the imaging of short Baseline Stereo, it is proposed that one kind is solved optimal based on random optimization
The new frame of parameter.
To solve the above problems, the present invention provides a kind of human hand based on the imaging of short Baseline Stereo and its object interaction tracking
Method, its main contents include：
(1) scene modeling；
(2) identification figure modeling；
(3) model evaluation；
(4) random optimization.
Wherein, described scene modeling, the paired stereo colour human hand image of given input, 1) use human hand characteristic
Storehouse is automatic and visually emulates human hand, using 22 bones carry the right hand simulation of skin speciality, meanwhile, it is imitative to this
The net filiform that the true right hand carries out similar blood vessels is filled, and common phase connects 1491 nodes, and the wrist has 26 freedom
Degree；Every hand is respectively by 27 state modulators：The position of 3 parameter characterization hands, 4 parameters are used for quartic element number and characterize hand
Body rotates and five fingers respectively use 4 parameter characterization angle of bend；2) for its scene of the interactive object of human hand
Modeling then uses 7 parameters, and 3 parameter characterization object's positions and 4 parameters are used for quartic element number and characterize object moment, the object
With 6 frees degree；To sum up, the human hand and its object model that all parameters are used to establish uniqueness can be managed with various dimensions array.
Described identification figure modeling, including identification division and appropriate response taxonomy and identification count.
Described identification division, contained according to uncertain entropy in information theory and contained much information in it is determined that the principle of entropy, is being examined
Simulation quality degree can be determined with this principle by looking into the extensibility of color in coloured image, be specially：For each pixel in image
P, the principal curvatures λ of neighbour B × B centered on it is calculated using local auto-correlation formula1And λ2, while without loss of generality, it is assumed that
λ1≥λ2And B=3, work as λ1And λ2Illustrate that color region is more unified during all in smaller value, work as λ1And λ2Illustrate when all larger
The turning of existing color, and work as λ1Compare λ2When much greater, then pixels illustrated p neighbour is fringe region, therefore identification can be used and breathed out
In this Corner Detection equation be defined as：
Ch=λ1·λ2-k(λ1+λ2)2 (1)
Wherein, k standards value is 0.04.
Described appropriateness response is established, and gives different responses for different color continuity, i.e., for as turning
Pixel region gives low-response, gives high response to the pixel region as edge, zero response is given to equalization region, in addition,
Relative measurement, vector (λ all are carried out to identification in every image1,λ2) magnitude d is defined as：
Logarithm is used to change ratio, then, calculates whole image d intermediate value md=d/2.
Described identification statistics, carries out recurrence calculating using activation primitive, identifies different degrees of response, it activates letter
Number is：
Then, defined variable a, to measure λ1And λ2Between difference, similarly use maA intermediate value is represented, is had：
According to formula (3) (4), the Correlation Identification degree of some pixel has：
Wherein threshold value wTDetermined by specific experiment value, according to formula (5), to the left and right of the paired three-dimensional human hand figure of input
The each pixel value of two figures takes c values, then can obtain two identification matching figure ClAnd Cr。
Described model evaluation, including color successive Assessment and scene modeling evaluation.
Described color successive Assessment, for the paired three-dimensional human hand image I of inputlAnd Ir, calculated according to formula (5)
Obtain its identification matching figure ClAnd Cr, it is assumed that the human hand scene of emulation is H, is existed comprising information such as position, direction, degree of crook
Interior, its three-dimensional coordinate is PH=(X, Y, Z), then can calculate projection view P both itlAnd Pr；In addition, using above- mentioned information, go back
Definable PlAnd PrContinuity s (the P of color point in viewl,Pr)：
By formula (6), global color continuity can be defined as：
Wherein, β is the parameter that control characteristic step number increases.
Described scene modeling evaluation, in the human hand image that inputs in pairs, if its emulating image can only correspond to wherein
One width input picture, the then emulation are removed, therefore, when a projection PlAppear in piece image, to cause another it is defeated
Entering image also has the projection P of matchingr, specifically, for certain pixel, if identical throwing can not be obtained in the range of 3 millimeters
Shadow, then the pixel is excluded, therefore scene evaluation takes the successional maximum of global color：
H*=arg max { SH(Il,Ir)} (7)
Wherein, SHIt is calculated by formula (7).
Described random optimization, pixel is considered as particle, based on particle cluster algorithm, each pixel keeps present situation, but makees
Retain an optimum position p for overall itsg, its value shared by population, then the speed v of particletWith position xtPass through formula (8)
(9) renewal is iterated, is had in each time t：
vt=K (vt-1+c1r1(pi-xt-1)+c2r2(pg-xt-1)) (8)
xt=xt-1+vt (9)
Wherein, K is compressibility factor constant, c1It is to perceive composition, c2Component of interaction, r1、r2It is univesral distribution between 01
Random sample value, in addition, c1+c2>4 permanent establishments, set c here1For 2.8, c2For 1.3, ψ=c is made1+c2, then haveTherefore the value of 27 final arguments is solved, the action of human hand can be successfully emulated and track itself and object
Interactive vestige.
Brief description of the drawings
Fig. 1 is a kind of system flow of human hand and its object interaction tracking based on the imaging of short Baseline Stereo of the present invention
Figure.
Fig. 2 is a kind of human hand detection of human hand and its object interaction tracking based on the imaging of short Baseline Stereo of the present invention
Figure.
Embodiment
It should be noted that in the case where not conflicting, the feature in embodiment and embodiment in the application can phase
Mutually combine, the present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.
Fig. 1 is the present invention a kind of human hand and its object interaction tracking flow chart based on the imaging of short Baseline Stereo.It is main
Want scene modeling；Identification figure modeling；Model evaluation；Random optimization.
Wherein, scene modeling, the paired stereo colour human hand image of given input, 1) it is automatic using human hand property data base
And human hand is visually emulated, using 22 bones carry the right hand simulation of skin speciality, meanwhile, to the emulation right hand
The net filiform for carrying out similar blood vessels is filled, and common phase connects 1491 nodes, and the wrist has 26 frees degree；Every
Hand is respectively by 27 state modulators：The position of 3 parameter characterization hands, 4 parameters be used for quartic element number characterize hand body rotate with
And five fingers respectively use 4 parameter characterization angle of bend；2) then make for its scene modeling with the interactive object of human hand
With 7 parameters, 3 parameter characterization object's positions and 4 parameters are used for quartic element number and characterize object moment, and the object has 6
The free degree；To sum up, the human hand and its object model that all parameters are used to establish uniqueness can be managed with various dimensions array.
Identification figure modeling, including identification division and appropriate response taxonomy and identification count.
Identification divides, and is contained according to uncertain entropy in information theory and contained much information in it is determined that the principle of entropy, colored checking
The extensibility of color can determine simulation quality degree with this principle in image, be specially：For each pixel p in image, use
Local auto-correlation formula calculates the principal curvatures λ of neighbour B × B centered on it1And λ2, while without loss of generality, it is assumed that λ1≥λ2
And B=3, work as λ1And λ2Illustrate that color region is more unified during all in smaller value, work as λ1And λ2There is color in explanation when all larger
Turning, and work as λ1Compare λ2When much greater, then pixels illustrated p neighbour is fringe region, therefore identification can use Harris angle
Point detection equation is defined as：
Ch=λ1·λ2-k(λ1+λ2)2 (1)
Wherein, k standards value is 0.04.
Appropriateness response is established, and gives different responses for different color continuity, i.e., for the pixel region as turning
Low-response is given in domain, gives high response to the pixel region as edge, zero response is given to equalization region, in addition, in every figure
Relative measurement, vector (λ all are carried out to identification as in1,λ2) magnitude d is defined as：
Logarithm is used to change ratio, then, calculates whole image d intermediate value md=d/2.
Identification counts, and carries out recurrence calculating using activation primitive, identifies different degrees of response, its activation primitive is：
Then, defined variable a, to measure λ1And λ2Between difference, similarly use maA intermediate value is represented, is had：
According to formula (3) (4), the Correlation Identification degree of some pixel has：
Wherein threshold value wTDetermined by specific experiment value, according to formula (5), to the left and right of the paired three-dimensional human hand figure of input
The each pixel value of two figures takes c values, then can obtain two identification matching figure ClAnd Cr。
Model evaluation, including color successive Assessment and scene modeling evaluation.
Color successive Assessment, for the paired three-dimensional human hand image I of inputlAnd Ir, it is calculated according to formula (5)
Identification matching figure ClAnd Cr, it is assumed that the human hand scene of emulation is H, including the information such as position, direction, degree of crook, thirdly
Dimension coordinate is PH=(X, Y, Z), then can calculate projection view P both itlAnd Pr；In addition, using above- mentioned information, definable is gone back
PlAnd PrContinuity s (the P of color point in viewl,Pr)：
By formula (6), global color continuity can be defined as：
Wherein, β is the parameter that control characteristic step number increases.
Scene modeling is evaluated, in the human hand image that inputs in pairs, if its emulating image can only correspond to wherein one it is defeated
Enter image, then the emulation is removed, therefore, when a projection PlAppear in piece image, to cause another width input picture
Also there is the projection P of matchingr, specifically, for certain pixel, should if identical projection can not be obtained in the range of 3 millimeters
Pixel is excluded, therefore scene evaluation takes the successional maximum of global color：
H*=arg max { SH(Il,Ir)} (7)
Wherein, SHIt is calculated by formula (7).
Random optimization, pixel is considered as particle, based on particle cluster algorithm, each pixel keeps present situation, but as overall
It retains an optimum position pg, its value shared by population, then the speed v of particletWith position xtEntered by formula (8) and (9)
Row iteration updates, and has in each time t：
Vx=K (vt-1+c1r1(pi-xt-1)+c2r2(pg-xt-1)) (8)
xt=xt-1+vt (9)
Wherein, K is compressibility factor constant, c1It is to perceive composition, c2Component of interaction, r1、r2It is univesral distribution between 01
Random sample value, in addition, c1+c2>4 permanent establishments, set c here1For 2.8, c2For 1.3, ψ=c is made1+c2, then haveTherefore the value of 27 final arguments is solved, the action of human hand can be successfully emulated and track itself and object
Interactive vestige.
Fig. 2 is a kind of human testing of human hand and its object interaction tracking based on the imaging of short Baseline Stereo of the present invention
Figure.As illustrated, it is observed that there is three row images, wherein the first from left row image is singlehanded tracking result, and a middle row image is
Human hand and pencil box interact tracking result, and a right row image is the tracking result of two human hands interactions, can be with from figure
Find out, be all divided into two per piece image, left side is actual human hand's posture, and right side is emulation posture, it can be seen that its model is imitated
The very close time of day of fruit.
For those skilled in the art, the present invention is not restricted to the details of above-described embodiment, in the essence without departing substantially from the present invention
In the case of refreshing and scope, the present invention can be realized with other concrete forms.In addition, those skilled in the art can be to this hair
Bright to carry out various changes and modification without departing from the spirit and scope of the present invention, these improvement and modification also should be regarded as the present invention's
Protection domain.Therefore, appended claims are intended to be construed to include preferred embodiment and fall into all changes of the scope of the invention
More and change.
Claims (10)
1. a kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo, it is characterised in that mainly including field
Scape models (one)；Identification figure modeling (two)；Model evaluation (three)；Random optimization (four).
2. based on the scene modeling (one) described in claims 1, it is characterised in that the paired stereo colour human hand of given input
Image, 1) use human hand property data base automatic and visually emulate human hand, carry out carrying skin to the right hand using 22 bones
The simulation of skin speciality, meanwhile, the net filiform that similar blood vessels are carried out to the emulation right hand is filled, and common phase connects 1491 and saved
Point, and the wrist has 26 frees degree；Every hand is respectively by 27 state modulators：The position of 3 parameter characterization hands, 4
Individual parameter is used for the sign hand body rotation of quartic element number and five fingers respectively use 4 parameter characterization angle of bend；2) for
Its scene modeling of the interactive object of human hand then uses 7 parameters, and 3 parameter characterization object's positions and 4 parameters are used for four
Dimension number characterizes object moment, and the object has 6 frees degree；To sum up, all parameters can be managed with various dimensions array to be used to build
Found unique human hand and its object model.
3. (two) are modeled based on the identification figure described in claims 1, it is characterised in that including identification division and appropriateness response
Classification and identification statistics.
4. based on the identification division described in claims 3, it is characterised in that contain information according to entropy is not known in information theory
Amount is more than the principle for determining entropy, and the extensibility of color can determine simulation quality degree with this principle in coloured image is checked, specifically
For：For each pixel p in image, neighbour B × B of the local auto-correlation formula calculating centered on it principal curvatures λ is used1
And λ2, while without loss of generality, it is assumed that λ1≥λ2And B=3, work as λ1And λ2Illustrate that color region is more united during all in smaller value
One, work as λ1And λ2Illustrate the turning of color occur when all larger, and work as λ1Compare λ2When much greater, then pixels illustrated p neighbour is
Fringe region, therefore identification can be defined as with Harris's Corner Detection equation：
Ch=λ1·λ2-k(λ1+λ2)2 (1)
Wherein, k standards value is 0.04.
5. established based on the appropriateness response described in claims 3, it is characterised in that given not for different color continuity
Same response, i.e., give low-response for the pixel region as turning, high response given to the pixel region as edge, to equilibrium
Zero response is given in region, in addition, all carrying out relative measurement, vector (λ to identification in every image1,λ2) magnitude d is defined as：
<mrow>
<mi>d</mi>
<mo>=</mo>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msqrt>
<mrow>
<msubsup>
<mi>&lambda;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>&lambda;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
</mrow>
</msqrt>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Logarithm is used to change ratio, then, calculates whole image d intermediate value md=d/2.
6. based on the identification statistics described in claims 3, it is characterised in that carry out recurrence calculating using activation primitive, know
Not different degrees of response, its activation primitive are：
<mrow>
<msub>
<mi>d</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mi>d</mi>
<mo>-</mo>
<msub>
<mi>m</mi>
<mi>d</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</msup>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
Then, defined variable a, to measure λ1And λ2Between difference, similarly use maA intermediate value is represented, is had：
<mrow>
<msub>
<mi>a</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mi>a</mi>
<mo>-</mo>
<msub>
<mi>m</mi>
<mi>a</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</msup>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
According to formula (3) (4), the Correlation Identification degree of some pixel has：
Wherein threshold value wTDetermined by specific experiment value, it is every to left and right two figure of the paired three-dimensional human hand figure of input according to formula (5)
Individual pixel value takes c values, then can obtain two identification matching figure ClAnd Cr。
7. based on the model evaluation (three) described in claims 1, it is characterised in that built including color successive Assessment and scene
Mould is evaluated.
8. based on the color successive Assessment described in claims 7, it is characterised in that for the paired three-dimensional human hand figure of input
As IlAnd Ir, its identification matching figure C is calculated according to formula (5)lAnd Cr, it is assumed that the human hand scene of emulation is H, includes position
Put, direction, including the information such as degree of crook, its three-dimensional coordinate is PH=(X, Y, Z), then can calculate projection view P both itl
And Pr；In addition, using above- mentioned information, definable P is gone backlAnd PrContinuity s (the P of color point in viewl,Pr)：
<mrow>
<mi>s</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>l</mi>
</msub>
<mo>,</mo>
<msub>
<mi>P</mi>
<mi>r</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mo>{</mo>
<msub>
<mi>C</mi>
<mi>l</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>l</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>C</mi>
<mi>r</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>r</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>}</mo>
<mo>&CenterDot;</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mi>&beta;</mi>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>I</mi>
<mi>l</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>l</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>,</mo>
<msub>
<mi>I</mi>
<mi>r</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>r</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mo>|</mo>
</mrow>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>6</mn>
<mo>)</mo>
</mrow>
</mrow>
By formula (6), global color continuity can be defined as：
<mrow>
<msub>
<mi>S</mi>
<mi>H</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>I</mi>
<mi>l</mi>
</msub>
<mo>,</mo>
<msub>
<mi>I</mi>
<mi>r</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>P</mi>
<mi>l</mi>
</msub>
<mo>&Element;</mo>
<msubsup>
<mi>R</mi>
<mi>l</mi>
<mi>H</mi>
</msubsup>
<mo>,</mo>
<msub>
<mi>P</mi>
<mi>r</mi>
</msub>
<mo>&Element;</mo>
<msubsup>
<mi>R</mi>
<mi>r</mi>
<mi>H</mi>
</msubsup>
</mrow>
</munder>
<mi>s</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>P</mi>
<mi>l</mi>
</msub>
<mo>,</mo>
<msub>
<mi>P</mi>
<mi>r</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, β is the parameter that control characteristic step number increases.
9. based on the scene modeling evaluation described in claims 8, it is characterised in that in the human hand image that inputs in pairs,
If its emulating image can only correspond to a wherein width input picture, emulation and be removed, therefore, when a projection PlAppear in one
In width image, to cause another width input picture that also there is the projection P of matchingr, specifically, for certain pixel, if at 3 millimeters
In the range of can not obtain identical projection, then the pixel is excluded, thus scene evaluation take the successional maximum of global color
Value：
H*=arg max { SH(Il,Ir)} (7)
Wherein, SHIt is calculated by formula (7).
10. based on the random optimization (four) described in claims 1, it is characterised in that pixel is considered as into particle, based on population
Algorithm, each pixel keeps present situation, but retains an optimum position p as overall itsg, its value shared by population, then grain
The speed v of sontWith position xtRenewal is iterated by formula (8) and (9), had in each time t：
vt=K (vt-1+c1r1(pi-xt-1)+c2r2(pg-xt-1)) (8)
xt=xt-1+vt (9)
Wherein, K is compressibility factor constant, c1It is to perceive composition, c2Component of interaction, r1、r2It is the random of univesral distribution between 01
Sample value, in addition, c1+c2>4 permanent establishments, set c here1For 2.8, c2For 1.3, ψ=c is made1+c2, then haveTherefore the value of 27 final arguments is solved, the action of human hand can be successfully emulated and track itself and object
Interactive vestige.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710632589.0A CN107392163A (en) | 2017-07-28 | 2017-07-28 | A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710632589.0A CN107392163A (en) | 2017-07-28 | 2017-07-28 | A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107392163A true CN107392163A (en) | 2017-11-24 |
Family
ID=60342061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710632589.0A Withdrawn CN107392163A (en) | 2017-07-28 | 2017-07-28 | A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107392163A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108387891A (en) * | 2018-03-21 | 2018-08-10 | 中国科学院电子学研究所 | The ULTRA-WIDEBAND RADAR received based on the hair of short baseline one two interferes two-dimensional localization tracking |
| CN109213322A (en) * | 2018-08-23 | 2019-01-15 | 深圳大学 | The method and system of gesture identification in a kind of virtual reality |
| CN109409255A (en) * | 2018-10-10 | 2019-03-01 | 长沙千博信息技术有限公司 | A kind of sign language scene generating method and device |
-
2017
- 2017-07-28 CN CN201710632589.0A patent/CN107392163A/en not_active Withdrawn
Non-Patent Citations (1)
| Title |
|---|
| PASCHALIS PANTELERIS等: "Back to RGB: 3D tracking of hands and hand-object interactions based on short-baseline stereo", 《HTTPS://ARXIV.ORG/ABS/1705.05301》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108387891A (en) * | 2018-03-21 | 2018-08-10 | 中国科学院电子学研究所 | The ULTRA-WIDEBAND RADAR received based on the hair of short baseline one two interferes two-dimensional localization tracking |
| CN108387891B (en) * | 2018-03-21 | 2021-08-31 | 中国科学院电子学研究所 | Ultra-wideband radar interference two-dimensional positioning and tracking method based on short baseline one-sending and two-receiving |
| CN109213322A (en) * | 2018-08-23 | 2019-01-15 | 深圳大学 | The method and system of gesture identification in a kind of virtual reality |
| CN109213322B (en) * | 2018-08-23 | 2021-05-04 | 深圳大学 | Method and system for gesture recognition in virtual reality |
| CN109409255A (en) * | 2018-10-10 | 2019-03-01 | 长沙千博信息技术有限公司 | A kind of sign language scene generating method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Oikonomidis et al. | Tracking the articulated motion of two strongly interacting hands | |
| CN104589356B (en) | The Dextrous Hand remote operating control method caught based on Kinect human hand movement | |
| Green et al. | Quantifying and recognizing human movement patterns from monocular video images-part i: a new framework for modeling human motion | |
| CN101807114B (en) | Natural interactive method based on three-dimensional gestures | |
| Lehrmann et al. | A non-parametric bayesian network prior of human pose | |
| Oprea et al. | A visually realistic grasping system for object manipulation and interaction in virtual reality environments | |
| CN111968165B (en) | Dynamic human body three-dimensional model complement method, device, equipment and medium | |
| CN102622766A (en) | Multi-objective optimization multi-lens human motion tracking method | |
| CN105912985A (en) | Human skeleton joint point behavior motion expression method based on energy function | |
| CN102184541A (en) | Multi-objective optimized human body motion tracking method | |
| CN107392163A (en) | A kind of human hand and its object interaction tracking based on the imaging of short Baseline Stereo | |
| Malik et al. | Simple and effective deep hand shape and pose regression from a single depth image | |
| Gonçalves et al. | A benchmark study on accuracy-controlled distance calculation between superellipsoid and superovoid contact geometries | |
| Endo et al. | Reconstructing individual hand models from motion capture data | |
| de La Gorce et al. | A variational approach to monocular hand-pose estimation | |
| Wu et al. | Log-gpis-mop: A unified representation for mapping, odometry, and planning | |
| Guo et al. | Inverse simulation: Reconstructing dynamic geometry of clothed humans via optimal control | |
| CN103839280B (en) | A kind of human body attitude tracking of view-based access control model information | |
| CN112149531B (en) | Human skeleton data modeling method in behavior recognition | |
| Bierbaum et al. | Robust shape recovery for sparse contact location and normal data from haptic exploration | |
| Koganti et al. | Real-time estimation of human-cloth topological relationship using depth sensor for robotic clothing assistance | |
| Li et al. | Fast vision-based pose estimation iterative algorithm | |
| CN110008873A (en) | Facial expression method for catching, system and equipment | |
| Wang et al. | Real-time control of 3D virtual human motion using a depth-sensing camera for agricultural machinery training | |
| Sorli et al. | Fine virtual manipulation with hands of different sizes |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20171124 |