CN106643563B - A kind of Table top type wide view-field three-D scanning means and method - Google Patents

Abstract

The invention discloses a kind of Table top type three-dimensional scanner and method, a kind of Table top type wide view-field three-D scanning means is made of multiple groups camera, laser MEMS galvanometer projector and precision rotation platform, and projector is arranged among multiple groups camera, is used for projecting structural optical；Multiple groups camera is symmetrically distributed in projector two sides, wherein symmetrical two cameras are one group.The present invention can cover the 3-D scanning from smaller size smaller to bulky objects, and object is smaller, and measurement accuracy is higher；One-touch scanning may be implemented in this method, obtains 360 degree of relatively good complete models.

Description

Desktop type large-view-field three-dimensional scanning device and method

Technical Field

The invention belongs to the field of optical detection, relates to an optical detection method for a three-dimensional profile, and particularly relates to a structured light three-dimensional measurement method with large depth of field and large field of view.

Background

Three-dimensional scanning has important application in the aspects of reverse engineering, industrial detection, cultural relic protection and the like, and along with the popularization of 3D printing, the three-dimensional scanning has wider application in the industries of reverse design, education training and the like. The existing three-dimensional scanners mainly aim at the professional fields of industrial detection and the like, and few scanners are provided for common consumers in industries such as vast reverse design and 3D printing education.

At present, the three-dimensional scanning generally adopts methods such as laser line scanning, white light structured light, stereoscopic vision and the like. The laser line scanning is developed most mature, the depth of field of measurement is large, but the efficiency is low, the splicing precision is low, the requirement on the precision of a moving device is high, and the cost is high. The white light structure is high in light efficiency, but its system calibration is difficult. The uncertainty of stereo matching in the stereo vision technology is low measurement precision. The phase shift profilometry in the white light structured light is combined with a binocular stereo vision method, so that the problems of difficult calibration of the phase shift profilometry and difficult stereo vision matching are solved. However, this approach requires encoding the light by a DLP projection device, whose measurement depth of field is limited by the projection depth of field; and simultaneously measuring the influence of the triangular relation included angle of the view field by binocular vision. In practical use, the sizes and the shapes of objects are different, and the measurement is invalid due to insufficient depth of field.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a desktop type three-dimensional scanning device, which adopts a laser MEMS galvanometer projection device to improve the depth of field of projection; capturing deformation fringe patterns in different view field ranges by adopting a plurality of groups of cameras, and improving the measurement range; and calibrating the position relation among the angle models by using a precision rotating table to splice the angle models into a complete model. The invention can cover three-dimensional scanning from a small volume to a large volume, and the smaller the measured object is, the higher the measurement precision is; the method can realize one-click scanning and obtain a better 360-degree complete model.

The invention aims to provide a three-dimensional scanning device for a desktop type large-view-field table.

The technical scheme is as follows:

the system consists of 4 to 10 cameras, 1 laser MEMS galvanometer projection device and a precise rotating platform; the laser MEMS galvanometer projection device is arranged in the middle and used for projecting structured light, the cameras are symmetrically arranged on two sides of the laser MEMS galvanometer projection device in pairs, and the distances among the symmetrically arranged cameras are L1, L2, L3, L4 and L5, wherein L1< L2< L3< L4< L5; the turret is placed within one or more fields of view of the camera. The turret is placed within one or more fields of view of the camera.

The laser MEMS galvanometer scanning projection device is used as a projection light source for three-dimensional scanning, the laser MEMS galvanometer scanning projection system is different from the traditional DLP, LCD and other projection systems, and is not provided with a projection lens. Because the light source is laser, the projection depth range is greatly improved.

The multiple groups of cameras are respectively used for shooting structured light in different fields of view to match the depth of field of the projection device, and the measuring range of the scanner is enlarged. The cost of the camera is much lower than that of the projection device, so that the cost of the equipment is not greatly increased, but the measurement range and the performance are greatly improved. Meanwhile, the basic conditions of binocular stereo vision are met, the visual fields of each group of cameras are overlapped as much as possible, and the included angle is generally 10-60 degrees, and preferably 15-45 degrees.

When the device is used for three-dimensional scanning, an object is placed on a rotating table and rotates for N times, wherein the rotating table rotates for α degrees each time, N is 360/α times, N is an integer, each time, a group of grating patterns or linear patterns are projected on the surface of the object by a laser MEMS galvanometer, the patterns on the surface of the object are shot by the camera, and the patterns are transmitted to a computer for three-dimensional point cloud calculation;

calibrating the position relation between the center of the precision rotating platform and the camera, and carrying out rigid transformation on the three-dimensional model reconstructed at each angle according to the calibration result to obtain a roughly spliced complete model; and performing optimized splicing by using ICP (inductively coupled plasma) to obtain a 360-degree three-dimensional digital model.

Advantageous effects

The invention uses the laser MEMS galvanometer projection device as a projection light source for three-dimensional structured light measurement, thereby greatly improving the projection depth of field; adopting a plurality of groups of camera designs to match the projection field of view; compared with the traditional three-dimensional scanner, the scanning range is increased by 3-5; the measurement accuracy of different sizes is different, and the accuracy of small objects is better; the precise rotary table is adopted for matching, so that the matching robustness is high, the effect is good, and the use difficulty of a user is reduced; the portable design of the turntable reduces the volume and weight of the product.

Drawings

FIG. 1 is a Gaussian laser beam propagation diagram;

FIG. 2 is a schematic diagram of a three-dimensional scanner structure arrangement;

FIG. 3 is a schematic view of a camera coordinate system and a turntable coordinate system;

wherein: 21-laser MEMS galvanometer projection device; 22-camera; a-range of small size objects; b-range of medium size objects; C-Large size object Range.

The specific implementation mode is as follows:

the calculation of the measurement range of the scanner is mainly determined by the projection device, the measurement range is matched by adopting a plurality of groups of cameras, and the configuration parameters of the scanner system are calculated as follows:

firstly, determining depth of field calculation of a laser MEMAS galvanometer projection system:

as shown in fig. 1, a gaussian laser beam emitted from a laser 11 passes through a collimating lens 12, and then enters an MEMS scanning galvanometer 13 and is reflected to the surface of an object. The focal plane of the Gaussian beam is L₀At, the depth of field is Δ L. Calculating the spot size and the maximum working distance L of the light beam after passing through the lens according to the ABCD matrix₂Size of spot omega₂Minimum working distance L₁The size of the light spot is omega₁Finally, L is determined by the following constraint of 1-1₁And L₂。

The depth of field of the laser MEMS galvanometer projection device can be determined according to the requirement of a measurement range. Second, designing the number of cameras to match the depth of field of the projection

The typical optical imaging system has a depth of field calculation formula of

In the formula: Δ L-Camera depth of field; Δ L₁、ΔL₂-the front and back depth of field of the imaging system; delta-the allowable circle diameter of dispersion, typically 2 times the size of the image element; f is the F number of the lens; l is the object distance. For example, F is 20mm, F is 2.0, and L is 500 mm; the pixel size is 7.5 μm, so δ is 0.015 mm. The depth of field DeltaL is calculated by a formula 3-2_C1＝75mm。

The system is compatible with measuring objects with different sizes, and has higher measuring precision for A group of small objects, the optical axes of each group of cameras are intersected at the center of the measuring view field, and the included angle is generally between 10 and 60 degrees, preferably between 15 and 45 degrees as shown in the following figure.

Mode of operation and data processing

The scanning mode of the desktop three-dimensional scanner designed by the invention is as follows, if the system comprises three groups of cameras:

1. designing a system:

selecting a measuring camera group such as group A, group B and group C according to the size of the measured object; each group of cameras corresponds to different measuring ranges, namely different sizes of measured objects.

a) Determining the measuring range, namely the depth range, of the laser MEMS galvanometer projection device according to the formula;

b) calculating the depth of field, namely the depth range, of the selected camera;

c) according to the method in the above inventive content, a suitable camera group is selected.

2. Rotary table calibration

As shown in the following figure, we use the rotation coordinate system as the world coordinate system and set a point P_w(x_w,y_w,z_w) Is an arbitrary point on the object to be measured, point P_θ(x_wθ,y_wθ,z_wθ) The coordinates corresponding to point P when the turntable is rotated counterclockwise about the axis by an angle θ, then they satisfy the relationship:

p＝R_zp_θ

wherein p is_w(x_w,y_w,z_w，1)^TAnd p_θ(x_wθ,y_wθ,z_wθ,1)^TIs P_w(x_w,y_w,z_w) And P_θ(x_wθ,y_wθ,z_wθ) Homogeneous coordinate in the turntable coordinate system, R_zIs a corresponding rotation matrix when the rotating platform rotates around the axis by an angle theta,

model point coordinate P obtained by reconstruction of actual person_c(x_c,y_c,z_c) Is in the camera coordinate system and corresponds to the point P in the world coordinate system_w(x_w,y_w,z_w) Satisfies the following conditions: p_c＝RP_w+T。

The coordinate transformation between the camera coordinates and the world coordinates obtained from the above formula can be represented by R (α, γ) and T (T ═ R (α, γ)_x,t_y,t_z)^Tα, gamma is the Euler angle of rotation around the x, y and z axes respectively, when coordinate transformation is performed successively with the x, y and z rotation,

let P_c(x_c,y_c,z_c) And P_w(x_w,y_w,z_w) Corresponding to a homogeneous coordinate of p_c(x_c,y_c,z_c1) and p_w(x_w,y_w,z_w1), then p_c＝Mp_w。

Wherein,since there is no requirement for the x-and y-axis directions by point cloud stitching with the rotating table coordinate system, γ can be set to 0, and the transformation matrix M can be simplified as:

thus, for the known point P in the camera coordinate system that we have reconstructed_c(x_c,y_c,z_c) The turntable rotates anticlockwise around the axis by an angle theta corresponding to a point P_cθ(x_cθ,y_cθ,z_cθ) They satisfy:

p_c＝MR_zM^-1p_cθ \*MERGEFORMAT(3-7)

wherein p is_c(x_c,y_c,z_c1) and p_cθ(x_cθ,y_cθ,z_cθ1) is P_c(x_c,y_c,z_c) And P_cθ(x_cθ,y_cθ,z_cθ) Corresponding homogeneous form.

Correctly placing the rotating table, and marking the chessboard grids at the center of the rotating table where the board is placed so that the object is in the centers of two images of a group of cameras; calculating the coordinate P of the corner point on the calibration plate under the camera coordinate system_c(x_c,y_c,z_c) Two calibration plate pictures are taken every 10 ° rotation between ± 60 ° and their world coordinates are calculated.

Thus, we obtained a series of p_c(x_c,y_c,z_c1) and p_cθ(x_cθ,y_cθ,z_cθ1), wherein

M is obtained from the above equation.

3. The object to be measured is placed on the rotating table, and is scanned once, that is, after the rotating table rotates α degrees each time, the rotating table is stationary for a period of time, the scanner completes model scanning of one angle, and performs N times of rotation measurement, where N is 360/α times (N is an integer), for example, α is 45 °, and N is 8.

4. According to the calibrated rotating table center result, carrying out rigid transformation on each angle, namely using to obtain a roughly spliced complete model;

the reconstructed known point P in the camera coordinate system_c(x_c,y_c,z_c) The angle α at which the turntable rotates anticlockwise about the axis corresponds to point P_cα(x_cα,y_cα,z_cα) They satisfy:

p_c＝MR_zM^-1p_cθ

wherein p is_c(x_c,y_c,z_c1) and P_cα(x_cα,y_cα,z_cα1) is P_c(x_c,y_c,z_c) And P_cα(x_cα,y_cα,z_cα) Corresponding homogeneous form. The matrix M is obtained from step 2 turntable calibration,

R_zis the corresponding rotation matrix when the turntable rotates around the axis by an angle α,α is the angle of rotation set manually in step 3.

Therefore, we use Pose ═ MR for the corresponding models_zM^-1And the result matrix is subjected to coordinate transformation to finish rough splicing of the point cloud model data.

5. Carrying out accurate splicing by using ICP (iterative Closed Point);

although the present invention has been described with reference to a preferred embodiment, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (3)

1. A desktop type large-view-field three-dimensional scanning device is characterized by comprising 10 cameras, 1 laser MEMS galvanometer projection device and a precise rotating platform, wherein the laser MEMS galvanometer projection device is arranged in the middle and used for projecting structured light, the cameras are symmetrically arranged on two sides of the laser MEMS galvanometer projection device in pairs, the distances among the symmetrically arranged cameras are L1, L2, L3, L4 and L5, L1< L2< L3< L4< L5, the rotating platform is arranged in one or more view field ranges of the cameras, an object is placed on the rotating platform to rotate N times during three-dimensional scanning, the object rotates α degrees every time, N is 360/α times, N is an integer, the laser MEMS galvanometer is used for projecting a group of patterns or linear patterns to the surface of the object once per rotation, the patterns on the surface of the object are transmitted to a shooting computer for three-dimensional calculation, the position relation between a calibration center and the cameras is calibrated, coarse grating angle reconstruction is carried out according to the grating angle, rigid splicing is carried out, and a point cloud model is obtained, and a complete three-dimensional calculation is carried out.

2. The scanning method of the desktop type large-field-of-view three-dimensional scanning device based on claim 1 is characterized in that: the scanner part adopts a laser MEMS galvanometer projection device to project structured light so as to improve the depth of field of projection; capturing deformation fringe patterns in different view field ranges by adopting a plurality of groups of cameras, and improving the measurement range; and calibrating the position relation among the angle models by using a precision rotating table, and splicing the scanning results of the three-dimensional scanner into a complete 360-degree model.

3. The scanning method of the desktop type large-field-of-view three-dimensional scanning device according to claim 2, wherein the calculation of the measurement range of the three-dimensional scanner is determined by the projection device, the measurement range is covered by a plurality of sets of cameras, and the configuration parameters of the scanner system are calculated as follows:

firstly, determining depth of field calculation of a laser MEMS galvanometer projection system:

after passing through a collimating lens, a Gaussian laser beam emitted by a laser is incident on an MEMS scanning galvanometer and then reflected to the surface of an object; the focal plane of the Gaussian beam is L₀The depth of field is Δ L; calculating the spot size and the maximum working distance L of the light beam after passing through the lens according to the ABCD matrix₂Size of spot omega₂Minimum working distance L₁The size of the light spot is omega₁Finally, L is determined by the following constraint of 1-1₁And L₂；

Determining the depth of field of the laser MEMS galvanometer projection device according to the requirement of a measurement range;

secondly, designing the number of cameras to match the depth of field of the projection:

the typical optical imaging system has a depth of field calculation formula of

In the formula: Δ L-Camera depth of field; Δ L₁-the front depth of field of the imaging system; Δ L₂-a back field of view of the imaging system; delta-the allowable circle diameter of dispersion, typically 2 times the size of the image element; f is the F number of the lens; l is the object distance.