CN103456016B - The body-sensing video camera net scaling method that a kind of visual angle is unrelated - Google Patents

Abstract

The present invention relates to the body-sensing video camera net scaling method that a kind of visual angle is unrelated, the method comprising the steps of: 1) coordinate system at the selected video camera place of the multiple cameras installed is the frame of reference, and all the other are non-referenced coordinate system；2) multiple cameras obtains the frame position information of same person simultaneously and shows；3) setting up human skeleton coordinate system, multiple cameras calculates spin matrix and the translation vector of the relative human skeleton coordinate system of each video camera respectively according to the frame position information detected；4) according to step 3) spin matrix that obtains and translation vector, calculate spin matrix and the translation vector of each non-referenced coordinate system relative datum coordinate system；5) according to step 4) spin matrix that obtains and translation vector, the frame position information MAP detected by non-referenced coordinate system place video camera is in the frame of reference.Compared with prior art, the present invention has the advantages such as simple to operate, real-time, the aid mark thing not needing specially to prepare other.

Description

The body-sensing video camera net scaling method that a kind of visual angle is unrelated

Technical field

The present invention relates to a kind of camera marking method, especially relate to the body-sensing video camera net scaling method that a kind of visual angle is unrelated.

Background technology

The demarcation of camera can be divided into two classes: a class is the demarcation of one camera；Another kind of is the global calibration of polyphaser.The demarcation of one camera refers to the Relation Parameters solved between every camera inside and outside parameter and camera.Conventional method is the ZhengyouZhang chessboard calibration method proposed and two standardizitions of RAC of Tsai proposition.Wherein, the chessboard calibration method that Zhang proposes, by by the summit of artificial labelling chessboard black and white grid as labelling point, stated accuracy is high, it has also become a kind of method of main flow.The global calibration of polyphaser refers in the data unification of all cameras to the overall frame of reference, namely obtains the relative coordinate between camera, namely translation vector T between camera coordinates system and spin matrix R.Main thought be find out joint detection between camera to the 3D labelling point spatial relationship to calculate between camera.Adopting the kinect point cloud detected at present is labelling point, calculates by ICP method.Major downside is that of said method is computationally intensive, it is necessary to extra auxiliary reference thing.

Summary of the invention

The purpose of the present invention is contemplated to overcome the defect that above-mentioned prior art exists and the body-sensing video camera net scaling method providing a kind of simple to operate, real-time visual angle unrelated.

The purpose of the present invention can be achieved through the following technical solutions:

The body-sensing video camera net scaling method that a kind of visual angle is unrelated, the method comprises the following steps:

1) coordinate system at the selected video camera place of the multiple cameras installed is the frame of reference, and all the other are non-referenced coordinate system；

2) multiple cameras obtains the frame position information of same person simultaneously and shows；

3) setting up human skeleton coordinate system, multiple cameras calculates spin matrix and the translation vector of the relative human skeleton coordinate system of each video camera respectively according to the frame position information detected；

4) according to step 3) spin matrix that obtains and translation vector, calculate spin matrix and the translation vector of each non-referenced coordinate system relative datum coordinate system；

5) according to step 4) spin matrix that obtains and translation vector, the frame position information MAP detected by non-referenced coordinate system place video camera is in the frame of reference, and shows, completes to demarcate.

Described step 3) in, based on the positional information of the left shoulder of human body, right shoulder and barycenter, build human skeleton coordinate system, particularly as follows:

With left shoulder to the direction of right shoulder for X-direction, it is Y direction with barycenter to the central point of right and left shoulders, determines Z-direction by the right-hand rule.

Described step 3) in, calculate the spin matrix of the relative human skeleton coordinate system of each video camera and be translated towards measurer body and be:

Obtain left shoulder in skeleton, right shoulder, barycenter at video camera S institute vector in a coordinate system and the vector in human skeleton coordinate system thereof, according to the below equation calculating video camera S-phase spin matrix to human skeleton coordinate systemAnd translation vector

^sP be a P video camera S expression vector corresponding in a coordinate system,^PeopleP is expression vector corresponding in human skeleton coordinate system for same point P,Represent the spin matrix of human skeleton coordinate system relative camera S place coordinate system,Represent the translation vector of human skeleton coordinate system relative camera S place coordinate system.

Described step 4) in, calculate the spin matrix of each non-referenced coordinate system relative datum coordinate system and be translated towards measurer body and be:

Expression vector corresponding in human skeleton coordinate system for some P^PeopleP is for being expressed as:

^BaseP be a P fiducial cameras expression vector corresponding in a coordinate system,Represent the spin matrix of the relative human skeleton coordinate system of fiducial cameras,Represent the fiducial cameras translation vector relative to human skeleton coordinate system,^Non-P be a P non-referenced video camera expression vector corresponding in a coordinate system,Represent the spin matrix of the relative human skeleton coordinate system of non-referenced video camera,Represent the non-referenced video camera translation vector relative to human skeleton coordinate system, then

The then spin matrix of non-referenced coordinate system relative datum coordinate systemThe translation vector of non-referenced coordinate system relative datum coordinate system

Compared with prior art, the present invention adopts the stable articulare of human body to be that reference point builds for the unrelated coordinate system in the visual angle of human skeleton, simple to operate, real-time, do not need specially to prepare other aid mark thing, and unrelated with the visual angle of each video camera.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the present invention；

Fig. 2 is the human skeleton figure that depth camera detects；

Fig. 3 is the human skeleton coordinate system schematic diagram built based on sane articulare.

Detailed description of the invention

Below in conjunction with the drawings and specific embodiments, the present invention is described in detail.The present embodiment is carried out premised on technical solution of the present invention, gives detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Embodiment

As shown in Figure 1, the body-sensing video camera net scaling method that a kind of visual angle is unrelated, with quickly sane demarcation body-sensing video camera net for target, it is proposed to the sane skeleton joint point of human body for reference point, build the human skeleton coordinate system that visual angle is unrelated, and adopt the method based on ordinate transform to realize.The method comprises the following steps:

In step S101, the coordinate system at the selected video camera place of the multiple cameras installed is the frame of reference, and all the other are non-referenced coordinate system, and the present embodiment is for 2, it is assumed that for A and B, and the coordinate system at selected video camera A place is the frame of reference；

In step s 102,2 video cameras obtain the frame position information of same person simultaneously and show；

In step s 103, setting up human skeleton coordinate system, 2 video cameras calculate the spin matrix of the relative human skeleton coordinate system of each video camera respectively according to the frame position information detected And translation vector

In step S104, according to the step S103 spin matrix obtained and translation vector, calculate spin matrix and the translation vector of each non-referenced coordinate system relative datum coordinate system；

In step S105, according to the step S104 spin matrix obtained and translation vector, the frame position information MAP detected at non-referenced coordinate system place video camera (referring here to B) is in the frame of reference (referring here to A), and shows, completes to demarcate.

To take on bone switch technology unrelated with the visual angle that barycenter is benchmark:

Depth camera kinect, under the assistance of software OpenNI, can effectively obtain 3D coordinate and the direction of 15 articulares of human body, and articulare is shown in Fig. 2.By observing and test discovery for a long time, in 15 articulares of acquisition, stable articulare mainly has 4, is left shoulder, right shoulder, head and barycenter respectively.Simultaneously, it was found that the upper part of the body torso portion of human body is all very stable, is difficult to change, the result that kinect detects also is the same.So, the present invention proposes to build human skeleton coordinate system based on left shoulder, right shoulder and barycenter.With left shoulder to the direction of right shoulder for X-direction, it is Y-direction with barycenter to the central point of right and left shoulders, determines Z-direction by the right-hand rule, be specifically shown in Fig. 3.

To calculate the spin matrix of the relative human skeleton coordinate system of video camera AAnd translation vectorFor example,

Wherein,^A1 P in P representation space camera A expression vector corresponding in a coordinate system,^PeopleP represents expression vector corresponding in human skeleton coordinate system for same point P,Represent the spin matrix of the relative camera A of human skeleton coordinate system,Representing that human skeleton coordinate system is relative to the translation vector at camera A, obtains by measuring, the expression of whole formula is meant that videos camera A institute in a coordinate system by the point in human skeleton coordinate system.

The translation vector on formula (1) the right is moved to left, formula (2) can be obtained:

Assuming that:P is it is known that then, and formula (2) is expressed as:

Wherein, willIt is expressed as:If^PeopleP=[100]^T, then:

If by right shoulder at the positional representation of human skeleton coordinate system being: (t, 0,0), then, can obtain:

Then can obtain $\left[\begin{array}{c}a\\ b\\ c\end{array}\right]=P/t\·---\left(4\right)$

In like manner, if^PeopleP=[010]^T, then:

If by barycenter at the positional representation of human skeleton coordinate system be (0 ,-s, 0), then:

Can obtain $\left[\begin{array}{c}d\\ e\\ f\end{array}\right]=P/(-s).---\left(5\right)$

By the right-hand rule, can obtainThe 3rd row be exactly the 1st row and the 2nd row apposition, then:

$\left[\begin{array}{c}g\\ h\\ i\end{array}\right]=\left[\begin{array}{c}b\×f-e\×c\\ c\×d-f\×a\\ a\×e-d\×b\end{array}\right]\·$

Then:

Due to, what be presently required is that current coordinate system is mapped to the spin matrix in human skeleton coordinate system, namely it is desirable that(representing the spin matrix of the relative human skeleton coordinate system of camera A), the feature of spin matrix can obtain:

Namely

In like manner need exist for solving be benchmark with human skeleton coordinate system translation vectorCan be obtained by formula (1)And due toTherefore:

Can obtain,

In like manner can obtain the spin matrix of the relative human skeleton coordinate system of video camera BAnd translation vector

Conversion between coordinate system:

The point that the non-referenced coordinate of 2 video cameras is detected is mapped in reference coordinate.In previous step, having solved spin matrix and the translation vector of relative human skeleton coordinate system, the purpose of this step is to solve for out spin matrix and the translation vector of video camera B relative camera A.Basic ideas are to obtain with human skeleton coordinate system for bridge.

Assuming that now with the coordinate system of the coordinate system of three coordinate system video camera A, the coordinate system of video camera B and human skeleton..It is expressed as by the ordinate transform of video camera A to people:

It is expressed as by the ordinate transform of video camera B to people:

Because:So, can obtainFurther, can obtain:So:

Finally, will solveWithPoint in non-referenced coordinate system B can be mapped in Basic Reference Coordinate System A, complete the demarcation of camera.

Claims (3)

1. the body-sensing video camera net scaling method that a visual angle is unrelated, it is characterised in that the method comprises the following steps:

1) coordinate system at the selected video camera place of the multiple cameras installed is the frame of reference, and all the other are non-referenced coordinate system；

2) multiple cameras obtains the frame position information of same person simultaneously and shows；

3) setting up human skeleton coordinate system, multiple cameras calculates spin matrix and the translation vector of the relative human skeleton coordinate system of each video camera respectively according to the frame position information detected；

Building based on the positional information of the left shoulder of human body, right shoulder and barycenter of human skeleton coordinate system, particularly as follows: with left shoulder to the direction of right shoulder for X-direction, be Y direction with barycenter to the central point of right and left shoulders, determine Z-direction by the right-hand rule；

4) according to step 3) spin matrix that obtains and translation vector, calculate spin matrix and the translation vector of each non-referenced coordinate system relative datum coordinate system；

5) according to step 4) spin matrix that obtains and translation vector, the frame position information MAP detected by non-referenced coordinate system place video camera is in the frame of reference, and shows, completes to demarcate.

2. the body-sensing video camera net scaling method that a kind of visual angle according to claim 1 is unrelated, it is characterised in that described step 3) in, calculate the spin matrix of the relative human skeleton coordinate system of each video camera and be translated towards measurer body and be:

Obtain left shoulder in skeleton, right shoulder, barycenter at video camera S institute vector in a coordinate system and the vector in human skeleton coordinate system thereof, according to the below equation calculating video camera S-phase spin matrix to human skeleton coordinate systemAnd translation vector

^SP be a P video camera S expression vector corresponding in a coordinate system,^PeopleP is expression vector corresponding in human skeleton coordinate system for same point P,Represent the spin matrix of human skeleton coordinate system relative camera S place coordinate system,Represent the translation vector of human skeleton coordinate system relative camera S place coordinate system.

3. the body-sensing video camera net scaling method that a kind of visual angle according to claim 2 is unrelated, it is characterised in that described step 4) in, calculate the spin matrix of each non-referenced coordinate system relative datum coordinate system and be translated towards measurer body and be:

Expression vector corresponding in human skeleton coordinate system for some P^PeopleP is for being expressed as:

^BaseP be a P fiducial cameras expression vector corresponding in a coordinate system,Represent the spin matrix of the relative human skeleton coordinate system of fiducial cameras,Represent the fiducial cameras translation vector relative to human skeleton coordinate system,^Non-P be a P non-referenced video camera expression vector corresponding in a coordinate system,Represent the spin matrix of the relative human skeleton coordinate system of non-referenced video camera,Represent the non-referenced video camera translation vector relative to human skeleton coordinate system, then

The then spin matrix of non-referenced coordinate system relative datum coordinate systemThe translation vector of non-referenced coordinate system relative datum coordinate system