CN1268892C - PSD based three-dimensional measuring method - Google Patents

Abstract

The present invention discloses a three-dimensional measurement method based on PSD. A camera device formed from a linear array PSD and a cylindrical lens is used for observing a spatial target point P to obtain the position signal of the spatial point P on the projection point u of the PSD camera; the position signal can uniquely define a spatial measurement plane. When a plurality of PSD cameras are simultaneously used, a plurality of projection points u obtained by the single spatial point P can define a plurality of measurement planes of the same number; the planes are non-parallel; the intersection point of the planes is the spatial point P. The position of the spatial point P can be figured out in a coordinate system defined by the planes under the condition that the equations of the planes can be determined. For realizing the three-dimensional positioning, firstly, the used PSD camera is calibrated to obtain the position relation of a reference coordinate system and the PSD projection point u. After the camera is calibrated, the three-dimensional positioning calculation is carried out. The present invention can realize the 3D real-time measurement of spatial point targets based on PSD devices.

Description

The method for three-dimensional measurement of position-based senser element PSD

Technical field

The present invention relates to position measurement technique, specifically a kind of method for three-dimensional measurement based on PSD.

Background technology

(the position sensing device: method for three-dimensional measurement Position Sensory Device) is purposes one of new and high technology very widely based on PSD.It is fast that PSD has response speed as a kind of location aware photoelectric device, characteristics such as resolution height; Can constitute point or body target are carried out the real-time measurement system of locus or pose, realize that the online 3D of noncontact measures.Pose and movement locus that such 3D measuring system can be applicable to automated arm and manipulating object detect, the Mechatronic Systems Performance Evaluation, locus error model data acquisition, position closed loop control, off-line or online distant operation, automatically control and motion planning, therefore various applications such as the deformation of heavy construction building and vibration detection have purposes very widely.

Traditional PSD position measurement all is one dimension or two dimension.When measuring 3D information, adopt the structured light pattern mostly, only be applicable to that among a small circle closely (tens to the hundreds of millimeter) uses.

Summary of the invention

The purpose of this invention is to provide a kind of method for three-dimensional measurement based on PSD that can improve the 3D measurement range and the distance of PSD device.

To achieve these goals, technical solution of the present invention is: use by linear array PSD to cooperate the cylinder camera lens to constitute camera apparatus observation space impact point P, obtain the position signalling of spatial point P at the subpoint u of PSD camera; The position signalling of subpoint u has determined to cross the position on spatial point P and the space measurement plane vertical with the camera lens bus; When a plurality of PSD cameras use simultaneously, can define same a plurality of measurement face by a plurality of subpoint u that single spatial point P obtained; These faces are non-parallel, and intersection point is this spatial point P; Under the confirmable condition of these plane equations, then the position of this spatial point P can be calculated under the coordinate system by the definition of described plane; For realizing three-dimensional localization, earlier use PSD camera to be demarcated, obtain the position relation of reference frame and PSD subpoint u, after camera calibration is finished, carry out three-dimensional localization again and calculate;

The 3D measuring system is demarcated the following method that adopts:

1) establishing the PSD camera coordinates is C, and the PSD imaging surface is I, u ₀Be defined as the initial point of imaging surface I, the direction of imaging surface I and PSD camera coordinates are that the Z axle of C is parallel, and W is a reference frame, so the pass of spatial point P and its subpoint u on PSD imaging surface I is: z/ (u-u ₀) du=y/f; Wherein, f: focal length, du: unit picture element length;

2) be that relational matrix between C, imaging surface I and these three coordinate systems of reference frame W is by the PSD camera coordinates:

$\left[\begin{array}{c}\mathrm{ut}\\ t\end{array}\right]=\left[\begin{array}{cccc}0& u_0& d_u& 0\\ 0& 1& 0& 0\end{array}\right]\left[\begin{array}{cccc}{R}_{11}& R_{12}& R_{13}& T_{14}\\ R_{21}& R_{22}& R_{23}& T_{24}\\ R_{31}& R_{32}& R_{33}& T_{34}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]---\left(1\right);$

Wherein, d _u=f/du;

3) establishing subpoint u can be unfolded as follows by formula (1) for the universal relation expression formula with space P:

m ₁₁x _w+m ₁₂y _w+m ₁₃z _w+m ₁₄-m ₂₁ux _w-m ₂₂uy _w-m ₂₃u2 _w＝u (3)；

4) matrix form of described formula (3) is: AM=U, and concrete expansion is:

$\left[\begin{array}{ccccccc}{x}_w& y_w& z_w& 1& -u{x}_w& -u{y}_w& -u{z}_w\end{array}\right]\left[\begin{array}{c}{m}_{11}\\ m_{12}\\ m_{13}\\ m_{14}\\ m_{21}\\ m_{22}\\ m_{23}\end{array}\right]=u---\left(4\right);$

5) when enough sampled point, described relational matrix M just can obtain with following formula:

M＝(A ^TA) ^-1A ^TAU (5)；

The 3D method for calculating and locating is:

1) can make plane equation into by described formula (3):

$\left[\begin{array}{cccccc}{m}_{11}& -m_{21}u& m_{12}& -m_{22}u& m_{13}& -m_{23}u\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\end{array}\right]=u-m_{14}---\left(6\right);$

2) after a plurality of PSD cameras provide a plurality of corresponding plane equations, under M determines situation, get final product the coordinate of spatial point P under W; Described formula (6) can be able to be write as matrix form BX=D,

Wherein $X={[x_w,y_w,z_w]}^T,B=\left[\begin{array}{c}{B}_1\\ B_2\\ B_3\end{array}\right],D=\left[\begin{array}{c}{D}_1\\ D_2\\ D_3\end{array}\right];B_i,D_{i,}i=\mathrm{1,2,3}......$ Calibrating parameters and measurement data formation for a plurality of measurement mechanisms; The expression formula of separating is:

X＝(B ^TB) ^-1B ^TBD (7)；

The number of known spatial point P will satisfy and separates Metzler matrix 〉=7 in the described scaling method, promptly has enough sampled points, and is evenly distributed on defined measurement space;

When a plurality of PSD camera measurement structure, can be write described formula (6) as matrix form BX=D by the 3D position, space of described formula (6) and (7) computer memory point P, wherein:

$X={[x_w,y_w,z_w]}^T,B=\left[\begin{array}{c}{B}_1\\ B_2\\ B_3\end{array}\right],D=\left[\begin{array}{c}{D}_1\\ D_2\\ D_3\end{array}\right];B_i,D_i,i=\mathrm{1,2,3}......$ Be a plurality of PSD

The calibrating parameters of camera and measurement data;

Described PSD camera is made of PSD device and cylinder camera lens; Measure described spatial point P, adopt at least 3 PSD cameras, can adopt following set-up mode: middle PSD camera lens bus and both sides PSD camera lens bus are mutual plumbness, and be vertical mutually to guarantee the measurement face that a plurality of PSD subpoint u image points are produced; Its visual angle is good equidistantly to be distributed as.

The present invention is based on the 3D measuring principle of PSD, compared with prior art have more following advantage:

1. adopted the 3D point target measurement pattern of orthogonal measuring hand-deliver remittance principle, can adopt the typical vision scaling method that measuring system is carried out parameter calibration, calculate (seeing system calibrating and method for calculating and locating) to finish 3D, be specially adapted on a large scale, remote (tens to hundreds of rice) uses.

2. utilize rapidity that the PSD position signalling obtains and unicity, high resolving power, can realize high-level efficiency, real-time and high resolving power that 3D measures.1. because PSD is the analog quantity photoelectric device, directly the position signalling of subpoint does not need to carry out the detection processing of image object, can have the speed of response of Millisecond.2. because the photosignal of PSD output is continuous, therefore has quite high resolution.These characteristics are by the character decision of device own.

Description of drawings

Fig. 1 is the used measuring system structural representation of the inventive method.

Fig. 2 is impact point of the present invention and measurement face mechanism that projection constituted thereof.

Fig. 3 is the position relation of spatial point of the present invention in each coordinate.

Embodiment

Below in conjunction with drawings and Examples in detail the present invention is described in detail.

A kind of method for three-dimensional measurement based on multi-thread battle array PSD camera of the present invention uses by linear array PSD to cooperate the cylinder camera lens to constitute camera apparatus observation space impact point P, obtains the position signalling of spatial point P at the subpoint u of PSD camera.This position signalling can space measurement plane of unique definition (seeing shown in Figure 2); When a plurality of PSD cameras use simultaneously, can define same a plurality of measurement face by a plurality of subpoint u that single spatial point P obtained; These faces are non-parallel, and its intersection point is exactly this spatial point P; Under the confirmable condition of these plane equations, then the position of this spatial point P can be calculated under the coordinate system by the definition of described plane; For realizing above-mentioned three-dimensional localization, earlier use PSD camera to be demarcated, obtain the position relation of reference frame and PSD subpoint u, after camera calibration is finished, carry out three-dimensional localization again and calculate.Measuring system constitutes as shown in Figure 1.

The key that the present invention is based in the method for three-dimensional measurement of linear array PSD is accurately to obtain target at the subpoint u of PSD position signalling.The position signalling of subpoint u has determined to cross the position (as shown in Figure 2) of spatial point P and the space plane vertical with the camera lens bus.Measure hand-deliver remittance positioning principle owing to adopt, therefore, the precision of u value has determined the bearing accuracy of measuring system.The position readings of subpoint u on the PSD camera can directly draw from analog computing circuit, and it is fast to have speed like this, characteristic of simple structure.

System calibrating of the present invention and method for calculating and locating are as follows:

According to the pin hole mapping principle, establishing the PSD camera coordinates is C, and the PSD imaging surface is I, u ₀Be defined as the initial point of imaging surface I, the direction of imaging surface I and PSD camera coordinates are that the Z axle of C is parallel, and W is a reference frame, so the pass of spatial point P and its subpoint u on PSD imaging surface I is: z/ (u-u ₀) du=y/f;

As shown in Figure 3, establishing the PSD camera coordinates is C, and image coordinate is I, u ₀Be the image coordinate system initial point, W is a reference frame.Wherein, f: focal length, du: unit picture element length; Relational matrix between PSD camera, PSD imaging surface I and the reference frame W is:

$\left[\begin{array}{c}\mathrm{ut}\\ t\end{array}\right]=\left[\begin{array}{cccc}0& u_0& d_u& 0\\ 0& 1& 0& 0\end{array}\right]\left[\begin{array}{cccc}{R}_{11}& R_{12}& R_{13}& T_{14}\\ R_{21}& R_{22}& R_{23}& T_{24}\\ R_{31}& R_{32}& R_{33}& T_{34}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]---\left(1\right);$

Wherein, d _u=f/du; Formula (1) can be rewritten into:

$\left[\begin{array}{c}\mathrm{ut}\\ t\end{array}\right]=\left[\begin{array}{cccc}{m}_{11}& m_{12}& m_{13}& m_{14}\\ m_{21}& m_{22}& m_{23}& m_{24}\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]---\left(2\right);$

Can get after the expansion:

m ₁₁x _w+m ₁₂y _w+m ₁₃z _w+m ₁₄-m ₂₁ux _w-m ₂₂uy _w-m ₂₃uz _w＝u (3)；

Formula (3) be exactly subpoint u for the universal relation expression formula of spatial point P.When relational matrix M was known, this was that parameter is the plane equation that is defined in georeferencing coordinate system W of u.In order to ask M, following formula can be rewritten into matrix form and be:

$\left[\begin{array}{ccccccc}{x}_w& y_w& z_w& 1& -u{x}_w& -u{y}_w& -u{z}_w\end{array}\right]\left[\begin{array}{c}{m}_{11}\\ m_{12}\\ m_{13}\\ m_{14}\\ m_{21}\\ m_{22}\\ m_{23}\end{array}\right]=u---\left(4\right);$

Its shorthand can be AM=U, and wherein M is a relational matrix.When enough sampled point, M just can obtain with following formula:

M＝(A ^TA) ^-1A ^TAU (5)；

After M determines, just can calculate the plane equation that defines by subpoint u with formula (3).

$\left[\begin{array}{cccccc}{m}_{11}& -m_{21}u& m_{12}& -m_{22}u& m_{13}& -m_{23}u\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\end{array}\right]=u-m_{14}---\left(6\right);$

After a plurality of PSD cameras provide a plurality of corresponding plane equations (present embodiment provides three plane equations by three PSD cameras), under M determined situation, the coordinate of spatial point P under W just can solve.Following formula can be write as BX=D.X=[x wherein _w, y _w, z _w] ^T, $B=\left[\begin{array}{c}{B}_1\\ B_2\\ B_3\end{array}\right],D=\left[\begin{array}{c}{D}_1\\ D_2\\ D_3\end{array}\right].$ B _i, D _i, i=1,2,3 is the calibrating parameters and the measurement data formation of three PSD cameras.So have:

X＝(B ^TB) ^-1B ^TBD (7)。

Concrete operations of the present invention are as follows:

1) PSD camera design.At first, press Fig. 2 principle design PSD camera, constitute by linear array PSD chip, cylinder camera lens and prime amplifier; Has projected nature to satisfy the imaging of spatial light target on PSD as Fig. 3 and Shi (2).Prime amplifier is used for the output signal of PSD is carried out pre-processing and amplifying.And according to this design at least three same PSD cameras; Described PSD camera 3 is made of PSD device 1 and cylinder camera lens 2.

2) many PSD measure state design.Place this three PSD cameras by Fig. 1 structure, and make middle PSD camera lens bus and both sides PSD camera lens bus be mutual plumbness.To guarantee that three PSD become the measurement face 4 that image point was produced vertical mutually.

3) measuring system constitutes.As shown in Figure 1, by at least three PSD cameras and its signal processing circuit, data handling machine constitutes complete many PSD three-dimensional measurement hardware system.Its signal processing circuit can be typical PSD position signalling computing circuit.

4) measuring system is demarcated.Measuring system demarcation and method for calculating and locating are described, with reference to formula (1)-(5), (number will satisfy and separates Metzler matrix 〉=7 known location point in given measurement range, generally get more than 20, be evenly distributed on defined measurement space) under the condition, obtain subpoint and the spatial point position relational matrix M under given reference frame.

5) 3D measures.After measuring system is demarcated, just can carry out the 3D position measurement of the arbitrary target point of this measurement space scope.Promptly when spatial point P had signal output on each PSD camera, data handling machine can obtain corresponding A/D value, this value corresponding space measurement face under the system calibrating meaning.When at least three PSD cameras adopted aforesaid measurement structure, just by formula (6) and (7) calculated the 3D position, space that P is ordered.And this position calculation speed is very high, can reach Millisecond.Thereby can satisfy real time position and monitor requirement with observing and controlling.

6) experimental result.Present embodiment uses three-dimensional platform (precision is 0.2mm) that described PSD camera spatial position measuring system is demarcated and detects, its sampling has real-time (tens bold and unconstrained seconds), spatial resolution reaches 1/5000, bearing accuracy reaches (the mean square deviation definition of millimeter level, measurement range 500*500*300mm), PSD camera and target range are 4m.

Adopt the inventive method, can realize the novel photoelectric 3D measuring system based on many PSD camera. The characteristics such as this system has resolution ratio height, fast response time, and antijamming capability is strong are specially adapted to big Scope, remote (tens to hundreds of rice) uses; Can realize high speed, high accuracy, high s/n ratio The 3D position probing. Compare with the stereoscopic vision that adopts ccd video camera, have good environmental adaptability, Certainty of measurement height and fast response time, low cost and other advantages. This require fast, online, high accuracy, The 3D position of circumstance complication or attitude measurement occasion and scientific research field especially have practical significance.

Claims (4)

1. the method for three-dimensional measurement of a position-based senser element PSD, it is characterized in that: use by linear array PSD to cooperate the cylinder camera lens to constitute camera apparatus observation space impact point P, obtain the position signalling of spatial point P at the subpoint u of PSD camera, the position signalling of subpoint u has determined to cross the position on spatial point P and the space measurement plane vertical with the camera lens bus; When a plurality of PSD cameras use simultaneously, can define same a plurality of measurement face by a plurality of subpoint u that single spatial point P obtained; These faces are non-parallel, and its intersection point is exactly this spatial point P; Under the confirmable condition of these plane equations, then the position of this spatial point P can be calculated under the coordinate system by the definition of described plane; For realizing three-dimensional localization, earlier use PSD camera to be demarcated, obtain the position relation of reference frame and PSD subpoint u, after camera calibration is finished, carry out three-dimensional localization again and calculate;

The 3D measuring system is demarcated the following method that adopts:

If the PSD camera coordinates is C, the PSD imaging surface is I, u ₀Be defined as the initial point of imaging surface I, the direction of imaging surface I and PSD camera coordinates are that the Z axle of C is parallel, and W is a reference frame, so the pass of spatial point P and its subpoint u on PSD imaging surface I is: z/ (u-u ₀) du=y/f; Wherein, f: focal length, du: unit picture element length;

By the PSD camera coordinates is that relational matrix between C, imaging surface I and these three coordinate systems of reference frame W is:

$\left[\begin{array}{c}\mathrm{ut}\\ t\end{array}\right]=\left[\begin{array}{cccc}0& u_0& d_u& 0\\ 0& 1& 0& 0\end{array}\right]\left[\begin{array}{cccc}{R}_{11}& R_{12}& R_{13}& T_{14}\\ R_{21}& R_{22}& R_{23}& T_{24}\\ R_{31}& R_{32}& R_{33}& T_{34}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]\·\·\·1);$

Wherein, d _u=f/du;

If subpoint u for the universal relation expression formula of space P can be by formula 1) be unfolded as follows:

m ₁₁x _w+m ₁₂y _w+m ₁₃z _w+m ₁₄-m ₂₁ux _w-m ₂₂uy _w-m ₂₃uz _w＝u 3)；

The matrix form of described formula (3) is: AM=U, and concrete expansion is:

$\left[\begin{array}{ccccccc}{x}_w& y_w& z_w& 1& -u{x}_w& -u{y}_w& {-\mathrm{uz}}_w\end{array}\right]\left[\begin{array}{c}{m}_{11}\\ m_{12}\\ m_{13}\\ m_{14}\\ m_{21}\\ m_{22}\\ m_{23}\end{array}\right]=u\·\·\·4);$

When enough sampled point, described relational matrix M just can obtain with following formula:

M＝(A ^TA) ^-1A ^TAU 5)；

The 3D method for calculating and locating is:

By described formula 3) can make plane equation into:

$\left[\begin{array}{ccc}{m}_{11}-m_{21}u& m_{12}-m_{22}u& m_{13}-m_{23}u\end{array}\right]\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\end{array}\right]=u-m_{14}\·\·\·6);$

After a plurality of PSD cameras provide a plurality of corresponding plane equations, under M determines situation, get final product the coordinate of spatial point P under W; Can be with described formula 6) can be write as matrix form BX=D,

X=[x wherein _w, y _w, z _w] ^T, $B=\left[\begin{array}{c}{B}_1\\ B_2\\ B_3\end{array}\right],$ $D=\left[\begin{array}{c}{D}_1\\ D_2\\ D_3\end{array}\right];$ B _i, D _i, i=1,2,3 is the calibrating parameters and the measurement data formation of a plurality of measurement mechanisms; The expression formula of separating is:

X＝(B ^TB) ^-1B ^T?BD 7)。

2. according to the method for three-dimensional measurement of the described position-based senser element of claim 1 PSD, it is characterized in that: the number of known spatial point P will satisfy and separates Metzler matrix 〉=7 in the described scaling method, promptly have enough sampled points, and be evenly distributed on defined measurement space.

3. according to the method for three-dimensional measurement of the described position-based senser element of claim 1 PSD, it is characterized in that: measure described spatial point P, adopt a plurality of PSD cameras, can adopt following set-up mode: middle PSD camera lens bus and both sides PSD camera lens bus are mutual plumbness, and be vertical mutually to guarantee the measurement face that a plurality of PSD subpoint u image points are produced.

4. according to the method for three-dimensional measurement of the described position-based senser element of claim 3 PSD, it is characterized in that: described a plurality of PSD camera quantity are at least 3, and its visual angle is with equidistant distribution.

Images