CN103472923B - A kind of three-dimensional virtual gesture selects the method for object scene - Google Patents

Abstract

The invention discloses a method for selecting a scene object by a three-dimensional virtual gesture, which comprises the following steps: (1) establishing a three-dimensional gesture model library; it is characterized in that it also includes the following steps: (2) determining the desktop equation of the placed object; (3) ) Initialization time T=0; (4) Predict the trajectory of the three-dimensional gesture, and use the position information H _T and H _{T+1 of the three-dimensional gesture T and T+1} to predict a straight line l: (5) Preliminarily select the candidate ( ₆ ) finalize the object set C2 to be selected; ( ₇ ) display the object to be selected, and highlight the candidate objects in the set C2; ( ₈ ) perform collision detection on the object to be selected , perform collision detection on each candidate object in the object set C ₂ in turn, if the collision detection is successful, then end, otherwise, T=T+1, T is the time, go to step (4). The invention realizes accurate selection of objects in the scene by three-dimensional gestures.

Description

A Method for Selecting Scene Objects with Three-dimensional Virtual Gestures

technical field

The invention relates to the field of three-dimensional virtual gestures, in particular to a method for selecting scene objects by three-dimensional virtual gestures.

Background technique

Selecting an object by pointing is the simplest action in reality. As long as a selection direction is given, the selected object can be judged by the intersection test. In the immersive environment, the viewpoint and the position of the virtual hand are often used to form the selection direction, while in the desktop environment, only the position of the mouse point is required. This technology can be traced back to the "put-that-there" interface developed by MIT in 1980, in which voice was used to confirm selection manipulation. Since then, a large number of selection techniques have been developed, which differ from these two design variables: the way the selection direction is formed and the range of selection (determining how many objects are likely to be selected per selection).

Since point selection basically requires no user's physical action, it can achieve good user performance just for selection. However, due to the limitation of the dimension, this pointing technique is not suitable for the object positioning operation. The user can only locate on an arc around the viewpoint, and cannot move the object along the selected direction.

Ray-Casting is the simplest point-and-select technique. The current selection direction is determined by the vector P attached to the virtual hand and the position h of the virtual hand: p(a)=h+ap, 0<a<+∞. In theory, this technique can select objects at any distance, but as the distance increases and the object size shrinks, the selection becomes more and more difficult. At the same time, due to the influence of hardware jitter, the error rate of selection is also gradually increased.

The "Spotlight" selection technology (Spotlight) replaces the original ray with a cone starting from the starting point of the selection vector. As long as an object falls within this cone, it will be selected, similar to the object being illuminated. This technique is especially beneficial for the selection of tiny objects, but at the same time, multiple objects fall into the cone, which will increase misoperations.

Aperture is an improved technology to Spotlfight, which allows users to interactively control the range of the projection when selecting objects with the cone. The projection direction of the cone is determined by the tracker of the head and the virtual hand, and the user can interactively adjust the distance between the virtual hand and the head to adjust the size of the cone to eliminate the ambiguity of the selection.

The idea of Image-Plane comes from a basic principle in graphics: that is, any object we see is a perspective projection on the plane behind the viewing pyramid, so only one point on the given projection plane can be selected. This technique, like the previous techniques, cannot change the distance between the selected object and the operator. Fishing-Ree1 makes up for this shortcoming. By adding additional input devices, such as buttons or sliders on the tracker, users can interactively Moves the selected object along the selection vector, but still degrades user performance due to the separate control of the six degrees of freedom of the object.

Contents of the invention

The technical solution to be solved by the present invention is to provide a method for selecting scene objects by three-dimensional virtual gestures, so as to realize accurate selection of objects in the scene by three-dimensional gestures.

The present invention adopts following technical scheme to realize the object of the invention:

A method for selecting a scene object with a three-dimensional virtual gesture, comprising the steps of:

(1) Establish a three-dimensional gesture model library;

It is characterized in that it also includes the following steps:

(2) Determine the desktop equation of the placed object;

(3) Initialization time T=0;

(4) Predict the trajectory of the three-dimensional gesture, and use the position information H _T and H _{T+1 of the three-dimensional gesture T and T+1} to predict a straight line l:

The equation of the line l is: $\left\{\begin{array}{c}x=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)x_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)x_{T+1}\\ \mathrm{the\; y}=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right){\mathrm{the\; y}}_T+\left(t-T\right)/\left(\left(T+1\right)-T\right){\mathrm{the\; y}}_{T+1}\\ z=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)z_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)z_{T+1}\end{array}\right.,$

t=0,1,2,Λ,n, where n is the number of frames to be predicted, where H _T =(x _T ,y _T ,z _T ),

H _T+1 ＝(x _T+1 ,y _T+1 ,z _T+1 );

(5) Preliminarily select the object set C ₁ to be selected;

(5.1) Obtain an intersection point o from the straight line l and the desktop Z;

(5.2) Set the threshold to be selected as R, where the threshold M _i is the coordinate of object i, and the number of objects left on the N desktop;

(5.3) Calculate the distance from each object to the intersection o in turn: The object with r _i < R is the candidate object, which is added to the set C ₁ of the candidate object;

(6) Finalize the object set C ₂ to be selected;

(6.1) Calculate the cosine value of the angle between the current position of the three-digit gesture of the candidate object C ₁ set and the vector formed by each candidate object and the normal vector of the hand, that is, cos<n,H _T+1 -M _i > , M _i is the coordinate of object i, n is the normal vector of the hand, and the normal vector of the hand is the vector perpendicular to the palm surface;

(6.2) Determine whether cos<n,H _T+1 -M _i > is greater than 0. If it is greater than 0, the object is a candidate object and added to the candidate object set C _2. This step is to put the object behind the three-dimensional gesture Candidate object exclusion;

(7) displaying the objects to be selected, and highlighting the candidate objects in the set C ₂ ;

(8) Perform collision detection on the objects to be selected, and perform collision detection on each candidate object in the object set C ₂ in turn, if the collision detection is successful, then end, otherwise, T=T+1, T is the time, go to step (4) .

As a further limitation to the technical solution, said step (2) includes the following steps:

(2.1) In the 3D scene, the 3D coordinates of point A in the lower left corner of the desktop and the vectors of the two sides intersecting the lower left corner of the desktop are known, that is, the two intersecting vectors on the desktop A known;

(2.2) From point A, the vector Get the plane equation of the desktop: Z=f(x,y,z).

Compared with the prior art, the advantages and positive effects of the present invention are: 1. The better selection technology mentioned in the background requires some equipment, such as adding buttons or sliding bars on the tracker, but the present invention does not. 2. The selection technology mentioned in the background is only based on the position of the object, but the present invention selects the object based on the status information of the gesture and the relative position of the hand and the object in the scene updated in real time. 3. The selection techniques mentioned in the background are basically long-distance selection. The probability of selecting an object is also affected by the distance and size of the object. For example, it is more difficult to select a smaller object in the scene. The present invention selects at a short distance, and only selects when the hand is relatively close to the object, so that the difficult selection of small objects does not occur, and the selection accuracy is high.

Description of drawings

Fig. 1 is a processing flowchart of the present invention.

detailed description

The present invention will be further described in detail below in conjunction with the accompanying drawings and preferred embodiments.

Referring to Figure 1, the 3D virtual gesture refers to a 3D gesture model that is generated by a computer based on computer vision technology, using tracking or reconstruction methods, and is consistent with the user's gesture. This is an existing technology, and the establishment of a 3D gesture model library is omitted here. step. The method of using 3D virtual gestures to select scene objects is as follows:

Step1.

Determine the desktop equation of the placed object, the three-dimensional coordinates of point A in the lower left corner of the desktop in the three-dimensional scene, and the vectors of the two sides intersecting the lower left corner of the desktop are known, that is, the two intersecting vectors on the desktop Known; from point A, the vector Get the plane equation of the desktop: Z=f(x,y,z).

Step2. Initialization time T=0;

Step3. Predict the motion curve of the three-dimensional hand, that is, the motion trajectory of the three-dimensional hand. Using the position information H _T and H _T+1 of the three-dimensional hands T and T+1 at the moment, predict a straight line l:

The equation of the line l is: $\left\{\begin{array}{c}x=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)x_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)x_{T+1}\\ \mathrm{the\; y}=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right){\mathrm{the\; y}}_T+\left(t-T\right)/\left(\left(T+1\right)-T\right){\mathrm{the\; y}}_{T+1}\\ z=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)z_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)z_{T+1}\end{array}\right.,$

t=0,1,2,Λ,n, where n is the number of frames to be predicted, where H _T =(x _T ,y _T ,z _T ),

H _T+1 = (x _T+1 , y _T+1 , z _T+1 ).

Step4.

Preliminary selection of object sets to be selected. Set the threshold to be selected as R, obtain an intersection point o from the straight line l and the desktop Z, and calculate the distance from each object to the intersection point o in turn; Objects with r _i <R are candidate objects, and are added to the set C ₁ of candidate objects. The threshold set in this paper is M _i is the coordinates of object i, and the number of objects left on the N desktop.

Step5.

Finally determine the set of objects to be selected, and sequentially calculate the cosine value of the angle between each object in the set of candidate objects C ₁ and the normal vector of the hand, that is, cos<n, H _T+1 -M _i >. The object of cos<n,H _T+1 -M _i >>0 is the candidate object, which is added to the set C ₂ of candidate objects. This step is to exclude candidates whose objects are behind the hand. Where cos<n,HT _+1- M _i >>0, the candidate object is in front of the hand at this time, cos<n,HT _+1- M _i ><0, the candidate object is behind the hand at this time. n is the normal vector of the hand.

Step6. Display the objects to be selected. Highlight the candidate objects in the set C ₂ .

Step7.

Perform collision detection on the object to be selected. Carry out collision detection _on each candidate object of C2 in turn, if the collision detection is successful, then end, otherwise, T=T+1, and go to Step3.

The undescribed technical features of the present invention can be realized by or using the prior art, and will not be repeated here. Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Those of ordinary skill in the art Changes, modifications, additions or substitutions made within the essential scope of the present invention shall also belong to the protection scope of the present invention.

Claims (2)

1. three-dimensional virtual gesture selects a method for object scene, comprises the steps:

(1) three-dimension gesture model storehouse is set up;

It is characterized in that, also comprise the steps:

(2) the desktop equation of put object is determined;

(3) initial runtime T=0;

(4) predict the movement locus of three-dimension gesture, utilize the positional information H in three-dimension gesture T and T+1 moment _tand H _t+1, prediction straight line l:

The equation of straight line l is: $\left\{\begin{array}{c}x=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)x_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)x_{T+1}\\ y=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)y_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)y_{T+1}\\ z=\left(t-\left(T+1\right)\right)/\left(T-\left(T+1\right)\right)z_T+\left(t-T\right)/\left(\left(T+1\right)-T\right)z_{T+1}\end{array}\right.,$

T=0,1,2 ..., n, wherein n is the frame number that will predict, wherein H _t=(x _t, y _t, z _t),

H _T+1＝(x _T+1,y _T+1,z _T+1)；

(5) preliminary selected object collection C to be selected ₁;

(5.1) an intersection point o is obtained by straight line l and desktop Z;

(5.2) setting threshold value to be chosen is R, wherein threshold value m _ifor the coordinate of object i, the number of the remaining object of N desktop;

(5.3) distance of each object to intersection point o is calculated successively: r _ithe object of < R is candidate's object, joins the set collection C of candidate's object ₁in;

(6) object collection C to be selected is finally determined ₂;

(6.1) calculated candidate object C successively ₁the cosine value of the angle between the normal vector of the vector that three gesture current locations of set and each candidate's object are formed and hand, i.e. cos<n, H _t+1-M _i>, M _ifor the coordinate of object i, n is the normal vector of hand, and the normal vector of hand is the vector perpendicular to palmar aspect;

(6.2) cos<n is judged, H _t+1-M _iwhether > is greater than 0, if be greater than 0, then this object is candidate's object, joins candidate's object collection C ₂in, this step is exactly got rid of by candidate's object of object after three-dimension gesture;

(7) show object to be selected, will C be gathered ₂in candidate's object carry out high light display and show;

(8) collision detection is carried out to object to be selected, successively to object collection C ₂each candidate's object carries out collision detection, if collision detection success, then terminate, otherwise T=T+1, T is the moment, go to step (4).

2. three-dimensional virtual gesture according to claim 1 selects the method for object scene, and it is characterized in that, described step (2) comprises the steps:

(2.1) select the three-dimensional coordinate of known desktop lower-left angle point A in three-dimensional scenic, and intersect at the known vector on two limits in the desktop lower left corner, two namely on desktop crossing vectors known;

(2.2) by an A, vector obtain the plane equation of desktop: Z=f (x, y, z).