CN102779344A - Registering block for space exchange and use method thereof - Google Patents

Abstract

The invention relates to a tag block for space exchange and a vision orientation registering method thereof. The tag block can reduce staff operation errors, improves registering precision, is compatible with various imaging devices and reduces cost of the tag block and orientation. The tag block is structurally characterized in that a cover plate is installed on a base, a vision tag is arranged on the cover plate, and the N-shaped tag block is arranged in the base; and the base is installed on a supporting plate, and imaging substances are filled in the N-shaped tag block. The method includes that (1) conducting imaging processing on the tag block through a perspective imaging device to obtain image body data; locating a middle point of the N-shaped tag block in a body data image, and using the coordinate of the middle point as the body data space coordinate of a tag point; (2) conducting automatic recognition and locating on the vision tag by using a three-dimensional vision locating system, calculating the coordinate of the middle point of the N-shaped tag block, and using the coordinate of the middle point as the vision space coordinate of the tag point; and (3) by means of an affine method or a singular value decomposition method, calculating a space alternation array of a body data image space and a vision space, and finishing registering of the body data space and the vision space.

Description

Spatial alternation is with registration block and method of application thereof

Technical field

The spatial alternation that the objective of the invention is to propose a kind of suitable vision localization is with registration block and be used for the volume data space and the method for visual space registration.

Background technology

Through visual space and volume data spatial registration, can utilize vision guide to accomplish process operation to interior of articles.Spatial registration method commonly used has three types:

1. use special framework to carry out registration.Object is installed in the registration framework, and the acquisition volume data are carried out registration through the characteristic of registration framework, and this method receives the restriction of frame size and mounting means, is difficult to adapt to dimension of object, shape variation.

2. utilize the geometric properties of object to carry out registration.As utilize the curved surface of body surface formation to carry out registration, and this mode calculated amount is very big, and the registration time is longer.

3. use registration block to carry out registration.At body surface fixed registration piece, be used for utilizing unique point to accomplish the registration in space at different space orientation and identification registration features point.When using the registration block registration, need to discern, locate the unique point that is used for spatial registration, but manual work is prone to introduce the personnel operation error when locating registration block.

For reducing the registration error of the third method, adopt the method for the automatic identification positioning of registration block usually, as adopting the spherical labels thing, handle the edge that image obtains ball, justify or oval match, and calculate sphere center position, the location of realizing registration block.Edge extracting needs great amount of calculation with the match of circle; Need the above treatment of picture of 2 width of cloth could in the volume data space, accomplish centre of sphere location; In order to carry out round reliably match, need to use large-sized spheroid, make that the registration block volume is bigger.Locator meams in the operating space to registration block is also more, like electromagnetic type, photo-electric, stereoscopic vision etc.In order to solve the compatibling problem of registration block and volume data imaging device, the registration block that electromagnetism and photoelectric method need use certain material to make.

Summary of the invention

The object of the invention is exactly for the automaticity that improves spatial registration, reduces the registration block cost, reduces the personnel operation error; Provide a kind of spatial alternation with registration block and vision localization method for registering thereof; It utilizes registration block in the registration space registration features point to be carried out automatic or semi-automatic location to reduce the personnel operation error in different treating; Improve registration accuracy, be easy to compatible various imaging device simultaneously, reduce the registration block cost.

For realizing above-mentioned purpose, the present invention adopts following technical scheme:

A kind of spatial alternation is used registration block, and it comprises visual indicia, cover plate, N shape mark, base.Visual indicia is pasted or is drawn on the cover board, and cover plate is installed on the base, in base, is provided with N shape mark; N shape mark is processed or is used the hollow tubular processing of having filled imaging pulvis or liquor by the material that is beneficial to the imaging of volume data collecting device.

Said visual indicia is similar to the tessellated black pattern piece of black and white for what have three public vertex at least, and these public vertex conllinear not.

Be full of the liquor or the pulvis that are used for the imaging of volume data imaging device in the said N shape mark.

Said cover plate one end is provided with recess, and said base is provided with tang, and recess and tang cooperatively interact.

A kind of vision localization method for registering that adopts spatial alternation with registration block,

(1) registration block is fixed on the operand, obtains the volume data of operand and registration block by perspective imaging equipment; Utilize two-layer faultage image in the volume data, the center of N shape mark in the registration block is positioned, the volume data volume coordinate of this centre coordinate as registration block;

(2) utilize the stereoscopic vision positioning system that visual indicia is discerned automatically, located, calculate the coordinate of N shape mark center in the registration block, with it as the coordinate of registration block in visual space;

(3) utilize affine method or SVD method, calculate the spatial alternation battle array of volume data space and visual space, the registration of perfect aspect data space and visual space.

In the said step 1), the volume data space-location method of registration block is:

Utilize perspective imaging equipment to obtain the faultage image of registration block; The N shape mark that comprises the pipeline formation that imaging pulvis or liquor are housed in the registration block; N shape is marked at has bright point-like imaging in the image, these imaging points can be with method detection and location such as Edge extraction.Note O is a volume data space initial point, and the summit, the lower left corner of N shape mark is designated as n ₁, three summits of along clockwise direction other are designated as n respectively ₂, n ₃, n ₄With the position as registration features point, the center of N shape mark, note should in the volume data space

Position coordinates is m _I(x, y, z), based on space vector relation, its coordinate can by

Calculate.

At faultage image I ₁In, can detect outward flange and cornerwise imaging point thereof of the N shape mark that obtains, be designated as a respectively ₁, b ₁, c ₁At faultage image I ₂The middle corresponding imaging point that obtains that detects is designated as a respectively ₂, b ₂, c ₂

Since on the N phenotypic marker, triangle a ₁n ₂b ₁Similar in appearance to triangle c ₁n ₄b ₁, i.e. Δ a ₁n ₂b ₁∝ Δ c ₁n ₄b ₁, its corresponding sides grow up to

Ratio, promptly $\left|\right|\frac{b_1{n}_2}{b_1{n}_4}\left|\right|=\left|\right|\frac{a_1{b}_1}{c_1{b}_1}\left|\right|=k,$ Wherein k is a constant.

Known N shape mark diagonal line design length is l, has

l＝‖n ₂n ₄‖＝‖b ₁n ₂‖+‖b ₁n ₄‖＝(1+k)‖b ₁n ₄‖＝(1+k)/k·‖b ₁n ₂‖

Because b ₁, b ₂Be positioned at vector

On, wherein, O is a volume data space initial point, know by the relation between space vector,

$\stackrel{\→}{{\mathrm{on}}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·k/{(1+k)}^{\·l}$

$\stackrel{\→}{{\mathrm{on}}_4}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_4}=\stackrel{\→}{{\mathrm{ob}}_1}-\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·1/{(1+k)}^{\·l}$

Therefore, in the volume data image, obtain b through Flame Image Process ₁, b ₂Coordinate after, can calculate the corresponding registration features point coordinate m of registration block _I

Said step 2) the visual space localization method of registration block is in:

Visual indicia is made up of 3 or more black pattern piece common point, utilizes the visual space location of this visual indicia to registration block; It is taken pictures to visual indicia with stereoscopic camera, utilizes image processing algorithm, extracts black pattern piece common point and calculates the three-dimensional coordinate of each common point; According to the difference of the geometric distributions characteristic of black pattern piece common point, can discern automatically and locate different registration block;

Note is any three black pattern piece common point A of conllinear not on the visual indicia, B, and the visual space coordinate of C is respectively

A _c(A _x, A _y, A _z), B _c(B _x, B _y, B _z), C _c(C _x, C _y, C _z), according to right-hand rule, have 1 D, the unit of making to

ABC directly in the plane; The visual coordinate that note D is ordered is D _c(D _x, D _y, D _z), this coordinate can by

With

Try to achieve.With A is initial point, sets up the reference frame coordinate system of visual indicia, note A, and B, C, D is at this reference frame

Under coordinate be respectively A _r(0,0,0), B _r(B _Rx, B _Ry, B _Rz), C _r(C _Rx, C _Ry, C _Rz), D _r(D _Rx, D _Ry, D _Rz), utilize affined transformation, try to achieve the transformation matrix that reference coordinate is tied to visual coordinate system and do ^CT _R,

${}^C{T}_R=\left[\begin{array}{cccc}{A}_x& B_x& C_x& D_x\\ A_y& B_y& C_y& D_y\\ A_z& B_z& C_z& D_z\\ 1& 1& 1& 1\end{array}\right]\·{\left[\begin{array}{cccc}0& B_{\mathrm{rx}}& C_{\mathrm{rx}}& D_{\mathrm{rx}}\\ 0& B_{\mathrm{ry}}& C_{\mathrm{ry}}& D_{\mathrm{ry}}\\ 0& B_{\mathrm{rz}}& C_{\mathrm{rz}}& D_{\mathrm{rz}}\\ 1& 1& 1& 1\end{array}\right]}^{-1}$

Can know the coordinate m of the center of N shape mark under visual indicia reference frame coordinate system by the design and installation parameter _r(x _r, y _r, z _r).The registration features point coordinate of registration block in visual space, the i.e. coordinate of the center of N shape mark under visual coordinate system

m _c(x _c, y _c, z _c), can by $\left[\begin{array}{c}{x}_c\\ y_c\\ z_c\\ 1\end{array}\right]={}^C{T}_R\left[\begin{array}{c}{x}_r\\ y_r\\ z_r\\ 1\end{array}\right]$ Calculate.

The method for registering of volume data space and visual space is in the said step 3):

Use four registration block, calculate the coordinate of N shape mark center in the registration block respectively according to the method for the invention, be the coordinate of registration features point, and be designated as m at volume data space and visual space _I1, m _I2, m _I3, m _I4And m _C1, m _C2, m _C3, m _C4, can be in the hope of the transformation matrix of volume data space to visual space ^CT _I, promptly

^CT _I=[m _C1m _C2m _C3m _C4] [m _I1m _I2m _I3m _I4] ^-1When using more registration block, can adopt the SVD method to find the solution the optimum solution of spatial alternation battle array.

The present invention is fixed on registration block on the processed object, utilizes the volume data image of fluoroscopic apparatus generating run object; Through Flame Image Process and calculating, accomplish the location of registration features point in the volume data space; Utilize stereoscopic camera to accomplish the location of registration features point in visual space; According to the elements of a fix of registration features point, realize the registration of volume data space to visual space in different spaces.

Registration block is made up of last/lower cover, visual indicia, N shape mark and framework installation pad, at the registration block outside surface, and the visual indicia that stickup or printing are made up of the tessellated black pattern piece of similar black and white.Be full of the liquor or the pulvis that are beneficial to the imaging of volume data imaging device in the N shape mark, can obtain blur-free imaging by imaging devices such as X-ray, CT, nuclear magnetic resonance.Visual indicia by the black pattern piece is formed can be by the automatic identification positioning of vision positioning system.

Registration block is fixedly connected on the operand, is carried out to picture by the volume data imaging device, obtains the volume data image.Utilize the imaging of N shape mark in 2 width of cloth faultage images, can accomplish the location of registration block in the volume data space.Visual indicia is by the automatic identification positioning of stereoscopic vision positioning system.After in volume data space and visual space, respectively registration block being positioned; Obtain the coordinate of registration features point; Carry out gained unique point coordinate in volume data space and the visual space corresponding; Utilize affine method or SVD method, calculate the spatial alternation battle array of volume data space and visual space, the registration of perfect aspect data space and visual space.This method may further comprise the steps:

(1) carries out imaging processing by perspective imaging equipment, obtain image volumetric data, utilize the imaging of N shape mark in 2 width of cloth faultage images, can in the volume data space, accomplish the location registration block;

(2) the automatic recognition visible sensation mark of stereoscopic vision positioning system and registration block positioned;

(3) coordinate of gained registration features point in volume data space and the visual space is carried out corresponding, utilize affine method or SVD method, calculate the spatial alternation battle array of volume data space and visual space, the registration of perfect aspect data space and visual space.

The invention has the beneficial effects as follows: the registration block process and assemble that N shape mark constitutes is simple, compact conformation, as long as can there be two width of cloth imaging of satisfying requirement of the present invention in N shape mark in volume data, can position it, is beneficial to the miniaturization that realizes registration block; Positioning action to N shape mark in volume data is simple, supports rapid registering; Through stereoscopic vision, can discern, locate different registration block automatically, improve the registration automaticity, avoid the personnel operation error, improve registration accuracy.

Description of drawings

Fig. 1: process flow diagram of the present invention;

Fig. 2: registration block structural representation;

Fig. 3: N shape is marked at the location in volume data space;

Fig. 4: the visual indicia example that the tessellated black pattern piece of black and white is formed that is similar to that is attached to the registration block outside surface.

Wherein, 1. cover plate, 2. visual indicia, 3. base, 4.N shape mark, 5. supporting plate.

Embodiment

Below in conjunction with accompanying drawing and enforcement the present invention is further specified.

Among Fig. 2, cover plate 1 is installed on the base 3, on cover plate 1, is provided with visual indicia 2, in base 3, is provided with N shape mark 4; Base 3 is installed on the supporting plate 5, in N shape mark 4, fills imaging substance.

Said visual indicia 2 is a density bullet.

Be full of the liquor or the pulvis that are used for the imaging of volume data imaging device in the said N shape mark 4.

Among Fig. 1, Fig. 3, Fig. 4, method of the present invention is:

(1) by perspective imaging equipment registration block is carried out imaging processing, obtain image volumetric data.Utilize the imaging of N shape mark in 2 width of cloth faultage images, can in the volume data space, position the center of N shape mark and with this centre coordinate as the coordinate of registration features point in the volume data space;

(2) utilize the stereoscopic vision positioning system that visual indicia is discerned automatically, located, and calculate the coordinate of N shape mark center, with it as the coordinate of registration features point in visual space;

(3) utilize affine method or SVD method, calculate the spatial alternation battle array of volume data space and visual space, the registration of perfect aspect data space and visual space.

(1) the volume data space orientation of registration block

Behind device scans such as CT, MR, can obtain the faultage image of N shape mark.If O is a volume data space initial point; N shape mark and faultage image I ₁Intersection point be a ₁, b ₁, c ₁With faultage image I ₂Intersection point be a ₂, b ₂, c ₂Note summit, the N shape mark lower left corner is n ₁, its summit is designated as n respectively along clockwise direction ₂, n ₃, n ₄,

By

Triangle a ₁n ₂b ₁Similar in appearance to triangle c ₁n ₄b ₁,

Have $\left|\right|\frac{b_1{n}_2}{b_1{n}_4}\left|\right|=\left|\right|\frac{a_1{b}_1}{c_1{b}_1}\left|\right|=k$ (corresponding sides grow up to ratio, and wherein k is a constant)

Known N shape mark diagonal line Design and Machining length is l, and b ₁Be positioned at straight line straight line n ₂n ₄On, according to geometric relationship, get l=‖ n ₂n ₄‖=‖ b ₁n ₂‖+‖ b ₁n ₄The ‖ b of ‖=(1+k) ₁n ₄‖=(1+k)/k ‖ b ₁n ₂‖

Because b ₁, b ₂Be positioned at vector

On, wherein, O is a volume data space initial point, know by the relation between space vector,

$\stackrel{\→}{{\mathrm{on}}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·k/{(1+k)}^{\·l}$

$\stackrel{\→}{{\mathrm{on}}_4}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_4}=\stackrel{\→}{{\mathrm{ob}}_1}-\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·1/{(1+k)}^{\·l}$

In the volume data image, obtain b through Flame Image Process ₁, b ₂Coordinate after, the note coordinate of registration features point in the volume data space is m _I(x, y, z), based on space vector relation, can by

$\stackrel{\→}{{\mathrm{om}}_I}=\stackrel{\→}{{\mathrm{on}}_2}+(\stackrel{\→}{{\mathrm{on}}_4}-\stackrel{\→}{{\mathrm{on}}_2})/2=\stackrel{\→}{{\mathrm{ob}}_1}+\frac{l}{2}\·\frac{k-1}{1+k}\·\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}$

Calculate the coordinate m of registration features point in the volume data space _I(x, y, z).

(2) visual space of registration block location

Visual indicia is made up of 3 or the black pattern piece common point that more is similar to black and white gridiron pattern intersection point, utilizes this visual indicia can realize the visual space location to registration block.With stereoscopic camera visual indicia is taken pictures, utilize image processing algorithm, extract black pattern piece common point and calculate the three-dimensional coordinate of each common point.Geometric distributions characteristic based on black pattern piece common point leaves like dot spacing, and recognition visible sensation marks such as angle are discerned automatically and located different registration block.

Note is any three black pattern piece common point A of conllinear not on the visual indicia, B, and the visual space coordinate of C is respectively

A _c(A _x, A _y, A _z), B _c(B _x, B _y, B _z), C _c(C _x, C _y, C _z), according to right-hand rule, there is 1 D, make vector of unit length Perpendicular to plane ABC.The visual coordinate that note D is ordered is D _c(D _x, D _y, D _z), then the D visual coordinate of ordering can by Try to achieve.With A is initial point, sets up the reference frame coordinate system of visual indicia, note

A, B, C, the coordinate of D under this reference frame is respectively A _r(0,0,0), B _r(B _Rx, B _Ry, B _Rz), C _r(C _Rx, C _Ry, C _Rz), D _r(D _Rx, D _Ry, D _Rz), utilize affined transformation, can to the transformation matrix that visual coordinate is do in the hope of reference frame ^CT _R,

${}^C{T}_R=\left[\begin{array}{cccc}{A}_x& B_x& C_x& D_x\\ A_y& B_y& C_y& D_y\\ A_z& B_z& C_z& D_z\\ 1& 1& 1& 1\end{array}\right]\·{\left[\begin{array}{cccc}0& B_{\mathrm{rx}}& C_{\mathrm{rx}}& D_{\mathrm{rx}}\\ 0& B_{\mathrm{ry}}& C_{\mathrm{ry}}& D_{\mathrm{ry}}\\ 0& B_{\mathrm{rz}}& C_{\mathrm{rz}}& D_{\mathrm{rz}}\\ 1& 1& 1& 1\end{array}\right]}^{-1}$

Can know that by the design and installation parameter coordinate of the center of N shape mark under visual indicia reference frame coordinate system is m _r(x _r, y _r, z _r).The registration features point coordinate of registration block in visual space, promptly the coordinate of the center of N shape mark under visual coordinate system is m _c(x _c, y _c, z _c), then by $\left[\begin{array}{c}{x}_c\\ y_c\\ z_c\\ 1\end{array}\right]={}^C{T}_R\left[\begin{array}{c}{x}_r\\ y_r\\ z_r\\ 1\end{array}\right]$ Calculate.

(3) registration of volume data space and visual space

Use four registration block, calculate the coordinate of N shape mark center in the registration block respectively according to the method for the invention,, and be designated as m as the position of registration features point at volume data space and visual space _I1, m _I2, m _I3, m _I4And m _C1, m _C2, m _C3, m _C4, can be in the hope of the transformation matrix of volume data space to visual space ^CT _I, promptly

^CT _I＝[m _C1?m _C2?m _C3?m _C4][m _I1?m _I2?m _I3?m _I4] ^-1。

When using more registration block, can utilize the SVD method to obtain the optimum solution of spatial alternation battle array.

Sign through above-mentioned mathematical way can comparatively accurately realize the location of registration block in volume data space and visual space, and the perfect aspect data space is to the registration of visual space.

(4) experiment and emulation

Utilize method among the present invention to try to achieve registration features point in the N shape mark and mainly receive the influence of image quality in the sterically defined precision of volume data, in faultage image to the positioning error of its imaging less than 1mm.

Use diameter repeatedly to test and the emulation experiment comparison as the N shape mark of 25mm*15mm as spherical labels thing and the 2mm caliber manufactured size of 10mm.In image space: the error of the spherical label in artificial location is 2.2mm; Spherical labels thing localization method positioning error is 0.82mm, and the inventive method positioning error is less than 0.5mm; In visual space: the spherical labels thing is difficult for realizing location automatically, and the positioning error that adopts semi-automatic localization method is 1.2mm, and adopting patent 201010105470.6 methods is 0.3mm to the positioning error of the said visual indicia of this patent.

Claims (8)

1. a spatial alternation is used tag block, it is characterized in that it comprises cover plate, and cover plate is installed on the base, on the cover board is provided with the visual indicia of pasting or drawing, and in base, is provided with N shape tag block; Base is installed on the supporting plate, is hollow tubular at N shape tag block, fills imaging substance in it.

2. spatial alternation as claimed in claim 1 is used tag block, it is characterized in that, said visual indicia is similar to the tessellated black pattern piece of black and white for what have three public vertex at least, and these public vertex conllinear not.

3. spatial alternation as claimed in claim 1 is used tag block, it is characterized in that, said imaging substance is liquor or the pulvis that is used for the imaging of volume data imaging device.

4. spatial alternation as claimed in claim 1 is used tag block, it is characterized in that, said cover plate one end is provided with recess, and said base is provided with tang, and recess and tang cooperatively interact.

5. a vision localization method for registering that adopts the described spatial alternation of claim 1 with tag block is characterized in that,

(1) registration block is fixed on the operand, obtains the volume data of operand and registration block by perspective imaging equipment; Utilize two-layer faultage image in the volume data, the center of N shape mark in the registration block is positioned, the volume data volume coordinate of this centre coordinate as registration block;

(2) utilize the stereoscopic vision positioning system that visual indicia is discerned automatically, located, calculate the coordinate of N shape mark center in the registration block, with it as the coordinate of registration block in visual space;

(3) utilize affine method or SVD method, calculate the spatial alternation battle array of volume data space and visual space, the registration of perfect aspect data space and visual space.

6. method as claimed in claim 5 is characterized in that, in the said step 1), the volume data space-location method of registration block is:

Utilize the faultage image of perspective imaging equipment acquisition registration block, comprise the N shape mark of the pipeline formation that imaging pulvis or liquor are housed in the registration block, N shape is marked at has bright point-like imaging in the image, and these imaging points are with method detection and location such as Edge extraction; Note O is a volume data space initial point, and the summit, the lower left corner of N shape mark is designated as n ₁, three summits of along clockwise direction other are designated as n respectively ₂, n ₃, n ₄With the position as registration features point, the center of N shape mark, this position coordinates of note is m in the volume data space _I(x, y, z), based on space vector relation, its coordinate by

Calculate;

At faultage image I ₁In, outward flange and cornerwise imaging point thereof of the N shape mark that detection obtains are designated as a respectively ₁, b ₁, c ₁At faultage image I ₂The middle corresponding imaging point that obtains that detects is designated as a respectively ₂, b ₂, c ₂

Since on the N phenotypic marker, triangle a ₁n ₂b ₁Similar in appearance to triangle c ₁n ₄b ₁, i.e. Δ a ₁n ₂b ₁∝ Δ c ₁n ₄b ₁, its corresponding sides grow up to ratio, promptly $\left|\right|\frac{b_1{n}_2}{b_1{n}_4}\left|\right|=\left|\right|\frac{a_1{b}_1}{c_1{b}_1}\left|\right|=k,$ Wherein k is a constant;

Known N shape mark diagonal line design length is l, has

l＝||n ₂n ₄‖＝‖b ₁n ₂‖+‖b ₁n ₄‖＝(1+k)‖b ₁n ₄‖＝(1+k)/k·‖b ₁n ₂‖

Because b ₁, b ₂Be positioned at vector

On, wherein, O is a volume data space initial point, know by the relation between space vector,

$\stackrel{\→}{{\mathrm{on}}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_2}=\stackrel{\→}{{\mathrm{ob}}_1}+\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·k/{(1+k)}^{\·l}$

$\stackrel{\→}{{\mathrm{on}}_4}=\stackrel{\→}{{\mathrm{ob}}_1}+\stackrel{\→}{b_1{n}_4}=\stackrel{\→}{{\mathrm{ob}}_1}-\frac{\stackrel{\→}{b_1{b}_2}}{\left|\right|b_1{b}_2\left|\right|}\·1/{(1+k)}^{\·l}$

Therefore, in the volume data image, obtain b through Flame Image Process ₁, b ₂Coordinate after, calculate the corresponding registration features point coordinate m of registration block _I

7. method as claimed in claim 5 is characterized in that, said step 2) in the visual space localization method of registration block be:

Visual indicia is made up of 3 or more black pattern piece common point, utilizes the visual space location of this visual indicia to registration block; It is taken pictures to visual indicia with stereoscopic camera, utilizes image processing algorithm, extracts black pattern piece common point and calculates the three-dimensional coordinate of each common point; According to the difference of the geometric distributions characteristic of black pattern piece common point, different registration block are discerned automatically and located;

Note is any three black pattern piece common point A of conllinear not on the visual indicia, B, and the visual space coordinate of C is respectively A _c(A _x, A _y, A _z), B _c(B _x, B _y, B _z), C _c(C _x, C _y, C _z), according to right-hand rule, there is 1 D, make vector of unit length

Perpendicular to plane ABC; The visual coordinate that note D is ordered is D _c(D _x, D _y, D _z), this coordinate by

With

Try to achieve; With A is initial point, sets up the reference frame coordinate system of visual indicia, note A, and B, C, the coordinate of D under this reference frame is respectively A _r(0,0,0), B _r(B _Rx, B _Ry, B _Rz), C _r(C _Rx, C _Ry, C _Rz), D _r(D _Rx, D _Ry, D _Rz), utilize affined transformation, try to achieve the transformation matrix that reference coordinate is tied to visual coordinate system and do ^CT _R,

${}^C{T}_R=\left[\begin{array}{cccc}{A}_x& B_x& C_x& D_x\\ A_y& B_y& C_y& D_y\\ A_z& B_z& C_z& D_z\\ 1& 1& 1& 1\end{array}\right]\·{\left[\begin{array}{cccc}0& B_{\mathrm{rx}}& C_{\mathrm{rx}}& D_{\mathrm{rx}}\\ 0& B_{\mathrm{ry}}& C_{\mathrm{ry}}& D_{\mathrm{ry}}\\ 0& B_{\mathrm{rz}}& C_{\mathrm{rz}}& D_{\mathrm{rz}}\\ 1& 1& 1& 1\end{array}\right]}^{-1}$

Can know the coordinate m of the center of N shape mark under visual indicia reference frame coordinate system by the design and installation parameter _r(x _r, y _r, z _r); The registration features point coordinate of registration block in visual space, i.e. the coordinate m of the center of N shape mark under visual coordinate system _c(x _c, y _c, z _c), by $\left[\begin{array}{c}{x}_c\\ y_c\\ z_c\\ 1\end{array}\right]={}^C{T}_R\left[\begin{array}{c}{x}_r\\ y_r\\ z_r\\ 1\end{array}\right]$ Calculate.

8. method as claimed in claim 5 is characterized in that, the method for registering of volume data space and visual space is in the said step 3):

Use four registration block, calculate the coordinate of N shape mark center in the registration block, be the coordinate of registration features point, and be designated as m at volume data space and visual space _I1, m _I2, m _I3, m _I4And m _C1, m _C2, m _C3, m _C4, try to achieve the transformation matrix of volume data space to visual space ^CT _I, promptly ^CT _I=[m _C1m _C2m _C3m _C4] [m _I1m _I2m _I3m _I4] ^-1When using more registration block, adopt the SVD method to find the solution the optimum solution of spatial alternation battle array.

Images