CN103294894A - Optimal computation method of rotating shaft in three-dimensional scanning system - Google Patents

Abstract

The invention is applicable to the technical field of electronics, and provides an optimal computation method of a rotating shaft in a three-dimensional (3D) scanning system. The method comprises the steps of obtaining an intersecting line of plane 3D data of multiple angles to serve as the initial rotating shaft of a rotating table, scanning a zero-angle 3D model, obtaining a first group of 3D data, controlling the rotating table to rotate a set angle, scanning and rotating a 3D model of the set angle, obtaining a second group of 3D data of the set angle, obtaining a coarse registration matrix through the set angle and the initial rotating shaft, computing a first optimal matrix through a fine-registration algorithm according to the first group of 3D data and the second group of 3D data of the set angle, scanning and rotating the 3D model, obtaining the 3D data of the set angle, computing a second optimal matrix, obtaining N optimal matrices, computing a registration error of each optimal matrix, and selecting the rotating shaft of the optimal matrix with the minimum registration error as the rotating shaft in the 3D scanning system. The optimal computation method has the advantage of computing the rotating shaft accurately.

Description

The optimized calculation method of the rotating shaft in a kind of 3 D scanning system

Technical field

The invention belongs to electronic applications, relate in particular to the optimized calculation method of the rotating shaft in a kind of 3 D scanning system.

Background technology

Existing 3D scanning system is merely able to the surperficial visual information in the sweep limit is scanned, in order to obtain the 3D model of complete object, usually need the scanner of 3D scanning system is moved, perhaps scanned object is rotated or moves, finish obtaining of scanned object different angles three-dimensional information.Because the position of the scanner of scanning system or the variation of scanned body position, each scan-data all is limited under different scanner coordinate systems or the different scanning object coordinate system, thereby causes the 3D data directly to be aligned in order in the unified coordinate system.In order to obtain the 3D information of complete object, need the 3D scan-data of these multi-angles be mated, in the existing 3D scanning system, matching process commonly used is that the labelling point is carried out in scanned surface, carry out registration operation by the gauge point that repeatedly scans the overlapping region that obtains, the subject matter that such operation brings: 1) binding mark is wasted time and energy, complex operation; 2) if the mark position of pasting is undesirable, may cause the registration failure; 3) disappearance of the three-dimensional information at binding mark place often needs artificially to repair by the cavity filling in later stage, and the 3D data are obtained imperfect.

In order to realize the autoregistration problem of multi-angle 3D scanning back data, 3D scan method based on accurate rotatable platform has appearred, object is placed on the turntable, by its rotational angle of computer control, the 3D information of realization object different angles is obtained, and then realizes the automatic splicing of multi-angle 3D data in conjunction with the rotational angle of turntable.The related subject matter of this method is the calculating of turntable rotating shaft, determine that namely rotating shaft is with respect to the locus of 3D scanister, method commonly used at present is the position of calculating its accurate axle center and axis by the approximating method of objects such as spheroid or plane, but the precision of this method depends on the accuracy of object putting position, if putting position has departed from the axle center of turntable, then be difficult to the accurate Calculation rotating shaft.

Summary of the invention

The purpose of the embodiment of the invention is to provide the optimized calculation method of the rotating shaft in a kind of 3 D scanning system, and the rotating shaft that is intended to solve prior art is calculated inaccurate, the problem that is difficult to the accurate Calculation rotating shaft.

The embodiment of the invention is achieved in that the optimized calculation method of the rotating shaft in a kind of 3 D scanning system, and described method comprises:

After planar object is placed on the turntable, rotate several times by turntable, obtain the many group plane three-dimensional data of this planar object under different rotational angles;

The plane three-dimensional data of these a plurality of angles of match is obtained the intersection of the plane three-dimensional data of a plurality of angles, and this intersection is the initial rotating shaft of turntable;

After this turntable scanned object was replaced with the 3D model, the 3D model of scanning zero degree obtained first group of three-dimensional data, and the control turntable rotates a set angle, and the 3D model of scanning rotation set angle obtains second group of three-dimensional data of this set angle;

Obtain thick registration matrix by set angle and this initial rotating shaft, slightly join the thick registration of second group of three-dimensional data after matrix is realized first group of three-dimensional data and rotated set angle according to this;

After thick registration is finished, according to second group of three-dimensional data of first group of three-dimensional data and set angle and calculate first by thin registration Algorithm and optimize matrix;

Continuation is rotated with set angle, and the 3D model of scanning rotation set angle obtains the 3rd group of 3 d scan data, calculates the second optimization matrix according to second group of three-dimensional data and the 3rd group of 3 d scan data by thin registration Algorithm;

Repeat and optimize the three-dimensional data that the matrix computations step is obtained N-2 optimization matrix and N-2 group set angle for N-2 time;

Calculate the registration error that each optimizes matrix, the rotating shaft of selecting the optimization matrix of registration error minimum is the rotating shaft in the 3 D scanning system.

Optionally, the registration error of described each optimization matrix of calculating specifically comprises:

The rotating shaft L1 that goes out according to first group of optimization matrix computations does mutual registration operation to remaining N group data, namely use first group of three-dimensional data of L1 registration, two groups of three-dimensional datas, second group of three-dimensional data, three groups of three-dimensional datas, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 group three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do mean value as the registration error of L1; Optimize matrix rotating shaft L2 with second and third group equally again and calculate first group of three-dimensional data, second group of three-dimensional data, second group of three-dimensional data, the 3rd group of three-dimensional data, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do the registration error that obtains L2 after average; Equally the rotating shaft of remaining optimization matrix is calculated the registration error of the rotating shaft of remaining optimization matrix.

Optionally, the scope of described set angle is:

10°—15°。

Optionally,, described N is the natural number more than or equal to 1.

The embodiment of the invention compared with prior art, beneficial effect is: the optimized calculation method of passing through not need accurately to put rotating shaft calculating object of reference of technical scheme of the present invention, at first a planar object roughly is placed in the turntable center, rotate N time by the control turntable, all turn to fixed angle (for example 10 °-20 °) from 0 degree at every turn, M organizes the 3D data on plane, obtains the roughly spatial parameter of rotating shaft by plane fitting.Near the turntable center, put an arbitrary objects then, and the roughly rotating shaft position that will before estimate carries out thick registration to this object, follow-up use closest approach iteration (ICP) is carried out meticulous registration operation, and be the optimization process that standard is carried out iteration with the registration error, minimum value after the whole registration error reaches convergence, can obtain accurate rotating shaft estimated value, and with this registration parameter as the different angles 3D data of follow-up 3D scanning process, realize full automatic registration operation, this invention has been avoided manually putting rotating shaft and has been calculated the operate miss that object causes, and can accurately obtain rotating shaft position and realize meticulous full-automatic registration operation.

Description of drawings

Fig. 1 is the process flow diagram of the optimization calculating side of the rotating shaft in a kind of 3 D scanning system of providing of the embodiment of the invention.

Embodiment

In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explaining the present invention, and be not used in restriction the present invention.

The specific embodiment of the invention provides the optimized calculation method of the rotating shaft in a kind of 3 D scanning system, and this method is finished by the 3D scanning system, and this method comprises as shown in Figure 1:

101, after planar object is placed on the turntable, the rotation several times obtain many groups plane three-dimensional data of a plurality of angles of this planar object;

102, the plane three-dimensional data of these a plurality of angles of match (2 angle or the angle more than 2) is obtained the intersection of the plane three-dimensional data of a plurality of angles, and this intersection is the initial rotating shaft of turntable;

103, after this turntable scanned object is replaced with the 3D model, the 3D model of scanning zero degree obtains first group of three-dimensional data, and the control turntable rotates a set angle, and the 3D model of scanning rotation set angle obtains second group of three-dimensional data of this set angle;

104, slightly joined matrix by set angle and this initial rotating shaft, slightly joined the thick registration that matrix is realized first group of three-dimensional data and second group of three-dimensional data of set angle according to this;

105, after thick registration is finished, according to second group of three-dimensional data of first group of three-dimensional data and set angle and calculate first by thin registration Algorithm and optimize matrix;

106, continue with the set angle rotation, the 3D model of scanning rotation set angle obtains the 3rd group of 3 d scan data, calculates the second optimization matrix according to second group of three-dimensional data and the 3rd group of 3 d scan data by thin registration Algorithm;

107, repeat 106N-2 time and obtain N-2 optimization matrix and N-2 group set angle three-dimensional data;

108, calculate the registration error that each optimizes matrix, the rotating shaft of selecting the optimization matrix of registration error minimum is the rotating shaft in the 3 D scanning system.

Above-mentioned continuation specifically can comprise with the set angle rotation: turns back to after the zero degree at every turn, and then the rotation set angle.

The scope of above-mentioned set angle is 10 °-15 °; Above-mentioned N can be for more than or equal to 1 natural number.

Need to prove, above-mentioned expression-form of slightly joining matrix specifically as shown in Equation 1:

$R\left(\mathrm{\θ}\right)=\left[\begin{array}{ccc}{u}^2+(v^2+w^2)\cos \mathrm{\θ}& \mathrm{uv}(1-\cos \mathrm{\θ})-w\sin \mathrm{\θ}& \mathrm{uw}(1-\cos \mathrm{\θ})+v\sin \mathrm{\θ}\\ \mathrm{uv}(1-\cos \mathrm{\θ})+w\sin \mathrm{\θ}& u^2+(v^2+w^2)\cos \mathrm{\θ}& \mathrm{vw}(1-\cos \mathrm{\θ})-u\sin \mathrm{\θ}\\ \mathrm{uw}(1-\cos \mathrm{\θ})-v\sin \mathrm{\θ}& \mathrm{vw}(1-\cos \mathrm{\θ})-u\sin \mathrm{\θ}& u^2+(v^2+w^2)\cos \mathrm{\θ}\\ 0& 0& 0\end{array}\right.$

$\left.\begin{array}{c}(a(v^2+w^2)-u(\mathrm{bv}+\mathrm{cw}))(1-\cos \mathrm{\θ})+(\mathrm{bw}-\mathrm{cv})\sin \mathrm{\θ}\\ (b(u^2+w^2)-v(\mathrm{au}+\mathrm{cw}))(1-\cos \mathrm{\θ})+(\mathrm{cw}-\mathrm{av})\sin \mathrm{\θ}\\ (c(u^2+v^2)-w(\mathrm{aw}+\mathrm{bv}))(1-\cos \mathrm{\θ})+(\mathrm{aw}+\mathrm{bv})\sin \mathrm{\θ}\\ 1\end{array}\right]$ Formula 1

Wherein θ is the anglec of rotation, and wherein a, b, c are the coordinate of initial rotating shaft L0 point A, and u, v, w are the tri-vector of initial rotating shaft vector V0, and wherein initial rotating shaft L0 is defined as: (c) (u, v w) form with a vectorial V0 for a, b by an A.

The above-mentioned first concrete manifestation form of optimizing matrix is as shown in Equation 2:

$R1\left(\mathrm{\θ}\right)=\left[\begin{array}{ccc}{u1}^2+({v1}^2+{w1}^2)\cos \mathrm{\θ}& u1v1(1-\cos \mathrm{\θ})-w1\sin \mathrm{\θ}& u1w1(1-\cos \mathrm{\θ})+v1\sin \mathrm{\θ}\\ u1v1(1-\cos \mathrm{\θ})+w1\sin \mathrm{\θ}& {u1}^2+({v1}^2+{w1}^2)\cos \mathrm{\θ}& v1w1(1-\cos \mathrm{\θ})-u1\sin \mathrm{\θ}\\ u1w1(1-\cos \mathrm{\θ})-v1\sin \mathrm{\θ}& v1w1(1-\cos \mathrm{\θ})-u1\sin \mathrm{\θ}& {u1}^2+({v1}^2+{w1}^2)\cos \mathrm{\θ}\\ 0& 0& 0\end{array}\right.$

$\left.\begin{array}{c}(a1({v1}^2+{w1}^2)-u1(b1v1+c1w1))(1-\cos \mathrm{\θ})+(b1w1-c1v1)\sin \mathrm{\θ}\\ (b1({u1}^2+{w1}^2)-v1(a1u1+b1w1))(1-\cos \mathrm{\θ})+(c1w1-a1v1)\sin \mathrm{\θ}\\ (c1({u1}^2+{v1}^2)-w1(a1w1+b1v1))(1-\cos \mathrm{\θ})+(a1w1-b1v1)\sin \mathrm{\θ}\\ 1\end{array}\right]$ Formula 2

Wherein, a1, b1, c1 are the coordinate of rotating shaft L1 point A1, and u1, v1, w1 are the tri-vector of rotating shaft vector V1, and its shaft L1 is defined as: by an A1 (a1, b1, c1) and a vectorial V1 (u1, v1 w1) form.

In like manner, can know that second optimizes matrix and N the expression-form of optimizing matrix, only be that the coordinate of rotating shaft is different with tri-vector in the expression-form of the second optimization matrix and N optimization matrix.Here repeat no more.

Above-mentioned thin registration Algorithm is specifically as follows the ICP algorithm, and specific implementation can be referring to the application of ICP algorithm in a cloud registration, Dai Jinglan, Chen Zhiyang, Ye Xiuzi, Chinese image graphics journal.

Optionally, above-mentioned 108 implementation method specifically can comprise:

The rotating shaft L1 that goes out according to first group of optimization matrix computations does mutual registration operation to remaining N group data, namely use first group of three-dimensional data of L1 registration, two groups of three-dimensional datas, second group of three-dimensional data, three groups of three-dimensional datas, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 group three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do mean value as the registration error of L1; Optimize matrix rotating shaft L2 with second and third group equally again and calculate first group of three-dimensional data, second group of three-dimensional data, second group of three-dimensional data, the 3rd group of three-dimensional data, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do the registration error that obtains L2 after average; Equally the rotating shaft of remaining optimization matrix is calculated the registration error of the rotating shaft of remaining optimization matrix.

Above-mentioned LN represents the rotating shaft of N group optimization matrix correspondence.

The optimized calculation method of passing through not need accurately to put rotating shaft calculating object of reference of technical scheme of the present invention, at first a planar object roughly is placed in the turntable center, rotate N time by the control turntable, all turn to fixed angle (for example 10 °-15 °) from 0 degree at every turn, M organizes the 3D data on plane, obtains the roughly spatial parameter of rotating shaft by plane fitting.Near the turntable center, put an arbitrary objects then, and the roughly rotating shaft position that will before estimate carries out thick registration to this object, follow-up use closest approach iteration (ICP, [1-2]) carry out meticulous registration operation, and be the optimization process that standard is carried out iteration with the registration error, minimum value after the whole registration error reaches convergence, can obtain accurate rotating shaft estimated value, and with this registration parameter as the different angles 3D data of follow-up 3D scanning process, realize full automatic registration operation, this invention has been avoided manually putting rotating shaft and has been calculated the operate miss that object causes, and can accurately obtain rotating shaft position and realize meticulous full-automatic registration operation.

Need to prove that above-mentioned scanner can be the instrument of single concrete scan function, can certainly be a plurality of instruments with scan function of combining, for example projector and digital camera.

The above only is preferred embodiment of the present invention, not in order to limiting the present invention, all any modifications of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.

Claims (4)

1. the optimized calculation method of the rotating shaft in the 3 D scanning system is characterized in that described method comprises:

After planar object is placed on the turntable, rotate some angles, carry out 3-D scanning simultaneously, obtain the plane three-dimensional data of a plurality of angles of the many groups of this planar object;

The plane three-dimensional data of these a plurality of angles of match is obtained the intersection of the plane three-dimensional data of a plurality of angles, and this intersection is the initial rotating shaft of turntable;

After this turntable scanned object was replaced with the 3D model, the 3D model of scanning zero degree obtained first group of three-dimensional data, and the control turntable rotates a set angle, and the 3D model of scanning rotation set angle obtains second group of three-dimensional data of this set angle;

Slightly joined matrix by set angle and this initial rotating shaft, slightly joined the thick registration that matrix is realized first group of three-dimensional data and second group of three-dimensional data according to this;

After thick registration is finished, according to first group of three-dimensional data and second group of three-dimensional data and calculate first by thin registration Algorithm and optimize matrix;

Continuation is rotated with set angle, and the 3D model of scanning rotation set angle obtains the 3rd group of 3 d scan data, calculates the second optimization matrix according to second group of three-dimensional data and the 3rd group of 3 d scan data by thin registration Algorithm;

Repeat and optimize the three-dimensional data that the matrix computations step is obtained N-2 optimization matrix and N-2 group set angle for N-2 time;

Calculate the registration error that each optimizes matrix, the rotating shaft of selecting the optimization matrix of registration error minimum is the rotating shaft in the 3 D scanning system.

2. method according to claim 1 is characterized in that, the registration error of each optimization matrix of described calculating specifically comprises:

The rotating shaft L1 that goes out according to first group of optimization matrix computations does mutual registration operation to remaining N group data, namely use first group of three-dimensional data of L1 registration, two groups of three-dimensional datas, second group of three-dimensional data, three groups of three-dimensional datas, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 group three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do mean value as the registration error of L1; Optimize matrix rotating shaft L2 with second and third group equally again and calculate first group of three-dimensional data, second group of three-dimensional data, second group of three-dimensional data, the 3rd group of three-dimensional data, the 3rd group of three-dimensional data, the 4th group of three-dimensional data, so that N-1 three-dimensional data, N organize three-dimensional data, obtain the accurate error of N assembly, do the registration error that obtains L2 after average; Equally the rotating shaft of remaining optimization matrix is calculated the registration error of the rotating shaft of remaining optimization matrix.

3. method according to claim 1 and 2 is characterized in that, the scope of described set angle is: 10 °-15 °.

4. method according to claim 1 and 2 is characterized in that, described N is the natural number more than or equal to 1.

Images