Method and device for scanning and three-dimensional reconstruction through line laser

Abstract

The invention discloses a three-dimensional reconstruction method and device based on line laser scanning, solves the problems that an existing binocular camera is poor in three-dimensional reconstruction precision, and line laser scanning is difficult in three-dimensional reconstruction, and achieves accurate and rapid three-dimensional reconstruction of a calibration object through a line laser scanner. Firstly, a high-precision rotary table and a laser scanning module are installed on the same plane, and a standard calibration cube is installed on the rotary table; then, a high-precision turntable encoder is controlled to collect depth data of a circle of a calibration object, the influence of acceleration and deceleration of a turntable is eliminated by removing overlapped point cloud parts at the beginning and the end, and outliers are eliminated through a filtering algorithm; carrying out the three-dimensional reconstruction of the calibration block through the average depth data of the point cloud, the scanning rotation angle, the geometric features of the calibration block, and the like, and obtaining a reconstructed model point cloud; and finally, plane fitting is performed on the reconstructed point cloud, and whether reconstruction succeeds or not is judged by judging whether normal vectors of the four surfaces are orthogonal or not. According to the method, the influence of acceleration and deceleration of the rotary table on reconstruction is reduced by eliminating the repeated point cloud part, the reconstruction precision is judged by adopting the geometrical characteristics of the reconstructed model, and the three-dimensional reconstruction precision and the three-dimensional reconstruction efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional reconstruction, and in particular to a method and a device for line laser scanning and three-dimensional reconstruction. Background Art

[0002] 3D reconstruction technology has been widely used in many fields, including virtual reality, augmented reality, robot navigation, cultural heritage protection, architectural design, etc. 3D reconstruction technology refers to the use of computer vision and image processing techniques to extract 3D geometric information from multiple images or data collected by sensors to generate or restore the corresponding 3D model or scene. Currently, 3D reconstruction is mostly done using binocular cameras or depth cameras. Traditional line laser scanning technology usually only provides a two-dimensional line projection information, and the application scenario is usually two-dimensional scanning, which cannot achieve 3D reconstruction.

[0003] At present, three-dimensional reconstruction mainly relies on binocular cameras, which are inefficient and cannot achieve high accuracy. A search of the literature on the prior art found that the Chinese invention patent publication number CN110349251A, named a three-dimensional reconstruction method and device based on binocular cameras, introduces a method for obtaining a three-dimensional network model by calculating the point cloud data of the common viewing area of ​​two target images through parallel arranged binocular cameras. However, this method requires preprocessing of the two images taken by the binocular camera in advance, which is less efficient and less accurate than a laser scanner. The Chinese invention patent publication number CN114413784A, named a line laser 3D contour scanning and reconstruction system and its control method, introduces a method and control method for three-dimensional contour measurement using a line laser. The method sets a slide assembly on the mounting frame, installs the line laser profiler on the slide, and drives the slide to move in a straight line by controlling the motor, so that the scanner is displaced relative to the object to be measured, thereby realizing three-dimensional reconstruction. However, this method can only three-dimensionally reconstruct part of the surface of the object to be measured, and cannot reconstruct the entire contour.

[0004] In summary, although existing binocular cameras can complete the task of 3D reconstruction, the images they capture often need to be preprocessed, resulting in low efficiency and lower accuracy than laser scanners. Current line laser scanners are usually suitable for scenarios such as 2D reconstruction. They extract 2D information of the object being measured by working with devices such as ball screws, but cannot achieve the goal of 3D reconstruction. Summary of the invention

[0005] The purpose of the present invention is to solve the problems of low three-dimensional reconstruction accuracy of binocular cameras and limited range of traditional line laser scanning, and to provide a method and device for three-dimensional reconstruction through line laser scanning.

[0006] The technical solution to achieve the purpose of the present invention is: a method and device for three-dimensional reconstruction by line laser scanning, including the following two stages:

[0007] Phase 1, the phase of obtaining reconstruction point data, includes the following steps:

[0008] Step 1.1, fix the bottom surface of the turntable and the laser scanning device in the same plane, so that the laser line scanning direction is parallel to the axis of the turntable;

[0009] Step 1.2, install a standard calibration cube at the center axis of the turntable, the side length of the cube is known;

[0010] Step 1.3, turn on the line laser scanner, determine the initial position under the turntable encoder and control the turntable to rotate more than one circle to collect point cloud data;

[0011] Step 1.4, remove the point cloud data whose start and end overlap according to the encoder position;

[0012] Step 1.5, use the statistical filtering algorithm to calculate the average distance between point clouds, remove outlier point clouds, and retain the point cloud coordinates of the calibrated part;

[0013] Step 1.6, convert the filtered point cloud coordinates into reconstructed point cloud coordinates according to the three-dimensional reconstruction algorithm.

[0014] Phase 2, judging the quality of 3D reconstruction, includes the following steps:

[0015] Step 2.1, intercepting the point cloud with the height value in the middle position and calculating the cross-sectional area of ​​the 3D reconstructed model;

[0016] Step 2.2, compare the cross-sectional area of ​​the cross-sectional cube calibration block of the reconstructed model with that of the calibration cube. If its area is larger than the cross-sectional area of ​​the calibration cube, reduce the parameters in the three-dimensional reconstruction algorithm. Otherwise, increase the parameters and repeat step 1.6. When the areas are the same, the reconstruction is considered successful.

[0017] Step 2.3, use the least squares method to perform plane fitting on the reconstructed point cloud coordinates to obtain the four fitting surfaces of the reconstructed calibration cube.

[0018] Step 2.4, calculate the normal vectors of the four faces according to the equations of the four fitting faces and determine whether the normal vectors are orthogonal to each other. If they are orthogonal to each other, it means that the three-dimensional reconstruction is successful.

[0019] Furthermore, in step 1.4, the overlapping parts of the start and end point clouds of the high-precision turntable are removed to reduce the impact of the turntable's start acceleration and end deceleration.

[0020] Furthermore, in step 1.6, the principle of the three-dimensional reconstruction algorithm is:

[0021]

[0022]

[0023] Z′=x

[0024] The coordinates of the scanned and filtered point cloud are P(x, y, z), the coordinates of the reconstructed point cloud are P′(x′, y′, z′), the average z value of the point cloud is h, and the side length of the cube calibration block is l.

[0025] Furthermore, in step 2.2, the method for adjusting the reconstruction parameters is:

[0026] g=s(l+h)-l 2

[0027]

[0028] The average z value of the point cloud is h, the side length of the cube calibration block is l, the area enclosed by the point cloud with the median z′ value after reconstruction is s(l+h), and g is the parameter judgment function.

[0029] Furthermore, in step 2.3, the method of calculating the fitting plane equation using the least squares method is:

[0030] AX=0

[0031] Where A is the n×4 fitting point cloud coordinate matrix, n is the number of fitting point clouds, and x is the 4×1 matrix composed of plane equation parameters, where the parameters are a, b, c, and d. Singular value decomposition is used to find the eigenvector corresponding to the last column of the singular vector matrix U, and the result of least squares fitting is obtained. The plane equation is ax+by+cz+d=0.

[0032] Furthermore, in step 2.4, the method for determining whether the normal vectors of the four fitting surfaces are orthogonal to each other is as follows:

[0033] α T β=0

[0034] α T γ=0

[0035] α T δ=0

[0036] β T γ=0

[0037] β T δ=0

[0038] γ T δ=0

[0039] Among them, α, β, γ, and δ are the normal vectors of the four fitting surfaces in step 2.4. If the normal vectors of the four fitting surfaces are orthogonal to each other, the three-dimensional reconstruction is completed.

[0040] Compared with the prior art, the present invention has the following significant advantages: (1) The present invention uses a line laser scanner for three-dimensional reconstruction, which improves the accuracy of three-dimensional reconstruction. (2) The present invention uses a turntable to perform three-dimensional reconstruction of the object to be measured, which makes the reconstruction area more complete than that of traditional devices. (3) The present invention judges the reconstruction quality by comparing the cross-sectional area of ​​the reconstructed model with the area of ​​the standard calibration block and the four fitting surface normal vectors, which improves the accuracy of reconstruction compared with other algorithms.

[0041] The present invention is further described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Flowchart of the 3D reconstruction method based on line laser scanning.

[0043] Figure 2 This is a schematic diagram of a three-dimensional reconstruction device based on line laser scanning. The measurement control module controls the scanning task of the laser scanning module and controls the rotation of the high-precision turntable to achieve three-dimensional reconstruction of the calibration module.

[0044] Figure 3 This is a three-dimensional reconstruction effect diagram based on line laser scanning. The depth point cloud distribution of the plane is obtained by scanning the object to be measured, and then the model is reproduced through the three-dimensional reconstruction algorithm. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] See also Figure 1 , is a flow chart of a three-dimensional reconstruction method based on line laser scanning in an embodiment of the present application. A three-dimensional reconstruction method and device based on line laser scanning in an embodiment of the present application comprises the following steps:

[0047] S1, obtain reconstruction point data, the specific process is:

[0048] S1.1, fix the turntable and the laser scanning module in the same plane, so that the laser line scanning direction is parallel to the turntable axis;

[0049] S1.2, install a standard calibration cube at the center axis of the turntable;

[0050] S1.3, turn on the line laser scanner, determine the initial position under the turntable encoder and control the turntable to rotate more than one circle to collect point cloud data;

[0051] S1.4, remove the point cloud data whose start and end overlap in combination with the encoder position information;

[0052] S1.5, using statistical filtering algorithm to remove outlier point clouds and retain the point cloud coordinates of the calibrated part;

[0053] S1.6, according to the 3D reconstruction algorithm, the filtered point cloud coordinates are converted into reconstructed point cloud coordinates. The principle of the algorithm is:

[0054]

[0055]

[0056] Z′=x

[0057] The coordinates of the scanned and filtered point cloud are P(x, y, z), the coordinates of the reconstructed point cloud are P′(x′, y′, z′), the average z value of the point cloud is h, and the side length of the cube calibration block is l.

[0058] S2, judging the reconstruction quality, the specific process is as follows:

[0059] S2.1, intercepting the point cloud with the height value in the middle position and calculating the cross-sectional area of ​​the 3D reconstructed model;

[0060] S2.2, compare the cross-sectional area of ​​the cross-sectional cube calibration block of the reconstructed model. If the area is larger than the cross-sectional area of ​​the calibration cube, reduce the parameters in the 3D reconstruction algorithm. Otherwise, increase the parameters and repeat S1.6. When the areas are the same, the reconstruction is considered successful. The specific judgment method is:

[0061] g=s(l+h)-l 2

[0062]

[0063] The average z value of the point cloud is h, the side length of the cube calibration block is l, the area enclosed by the point cloud with the median z′ value after reconstruction is s(l+h), and 9 is the parameter judgment function.

[0064] S2.3, the least square method is used to perform plane fitting on the reconstructed point cloud coordinates to obtain four fitting surfaces of the reconstructed calibration cube. The fitting method is:

[0065] AX=0

[0066] Where A is the n×4 fitting point cloud coordinate matrix, n is the number of fitting point clouds, and X is the 4×1 matrix composed of plane equation parameters, where the parameters are a, b, c, and d. Singular value decomposition is used to find the eigenvector corresponding to the last column of the singular vector matrix U, and the result of least squares fitting is obtained. The plane equation is ax+by+cz+d=0.

[0067] S2.4, calculate the normal vectors of the four surfaces according to the equations of the four fitting surfaces and determine whether the normal vectors are orthogonal to each other. If they are orthogonal to each other, it means that the 3D reconstruction is successful. The method for determining whether the normal vectors of the four fitting planes are orthogonal is as follows:

[0068] α T β=0

[0069] α T γ=0

[0070] α T δ=0

[0071] β T γ=0

[0072] β T δ=0

[0073] γ T δ=0

[0074] Among them, α, β, γ, and δ are the normal vectors of the four fitting surfaces in S2.3. If the normal vectors of the four fitting surfaces are orthogonal to each other, it means that the three-dimensional reconstruction is completed.

[0075] See also Figure 3 , is a three-dimensional reconstruction effect diagram of an embodiment of the present application.

[0076] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method and device for three-dimensional reconstruction by line laser scanning, characterized in that: It includes the following two stages: Phase 1, obtaining reconstruction point data, includes the following steps: Step 1.1, fix the bottom surface of the turntable and the laser scanning device in the same plane, so that the laser line scanning direction is parallel to the axis of the turntable; Step 1.2, install a standard calibration cube at the center axis of the turntable; Step 1.3, turn on the line laser scanner, determine the initial position under the turntable encoder and control the turntable to rotate more than one circle to collect point cloud data; Step 1.4, remove the point cloud data whose start and end overlap according to the encoder position; Step 1.5, use the statistical filtering algorithm to calculate the average distance between point clouds, remove outlier point clouds, and retain the point cloud coordinates of the calibrated part; Step 1.6, converting the filtered point cloud coordinates into reconstructed point cloud coordinates according to the three-dimensional reconstruction algorithm; Stage 2, judging the quality of 3D reconstruction, includes the following steps: Step 2.1, intercepting the point cloud with the height value in the middle position and calculating the cross-sectional area of ​​the 3D reconstructed model; Step 2.2, compare the cross-sectional area of ​​the cross-sectional cube calibration block of the reconstructed model with that of the calibration cube. If its area is larger than the cross-sectional area of ​​the calibration cube, reduce the parameters in the three-dimensional reconstruction algorithm. Otherwise, increase the parameters and repeat step 1.

6. When the areas are the same, the reconstruction is considered successful. Step 2.3, use the least squares method to perform plane fitting on the reconstructed point cloud coordinates to obtain the four fitting surfaces of the reconstructed calibration cube. Step 2.4, calculate the normal vectors of the four faces according to the equations of the four fitting faces and determine whether the normal vectors are orthogonal to each other. If they are orthogonal to each other, it means that the three-dimensional reconstruction is successful.

2. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In the aforementioned stage one and stage two, the three-dimensional reconstruction device includes a laser scanning module, a high-precision turntable, a reconstruction calibration module and a measurement control module.

3. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 1.2, the upper plane of the line laser scanner and the upper plane of the cube calibration block are adjusted to be perpendicular to the ground by placing a level, thereby ensuring that the line laser is parallel to the axis of the calibration block.

4. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 1.3, the specific angle of controlling the turntable to rotate one circle is empirically in the range of 370° to 380°.

5. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 1.4, the overlapped parts of the high-precision turntable start and end point clouds are removed to reduce the impact of the turntable's start acceleration and end deceleration.

6. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 1.6, the principle of the 3D reconstruction algorithm is: Z'=x The coordinates of the scanned and filtered point cloud are P(x, y, z), the coordinates of the reconstructed point cloud are P′(x′, y′, z′), the average z value of the point cloud is h, and the side length of the cube calibration block is l.

7. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 2.1, the point cloud with a middle height is the point cloud coordinate with the median z′ value in the reconstructed point cloud coordinates P′(x′, y′, z′).

8. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 2.2, g=s(l+h)-l 2 The average z value of the point cloud is h, the side length of the cube calibration block is l, the area enclosed by the point cloud with the median z′ value after reconstruction is s(l+h), and g is the parameter judgment function.

9. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 2.3 AX=0 Where A is the n×4 fitting point cloud coordinate matrix, n is the number of fitting point clouds, and X is the 4×1 matrix composed of plane equation parameters, where the parameters are a, b, c, and d. Singular value decomposition is used to find the eigenvector corresponding to the last column of the singular vector matrix U, and the result of least squares fitting is obtained. The plane equation is ax+by+cz+d=0.

10. The method and device for three-dimensional reconstruction by line laser scanning according to claim 1, characterized in that: In step 2.4 α T ·β＝0 a T ·γ=0 a T ·δ=0 b T ·γ=0 b T ·δ=0 c T ·δ=0 Among them, α, β, γ, and δ are the normal vectors of the four fitting surfaces in step 2.

4. If the normal vectors of the four fitting surfaces are orthogonal to each other, the three-dimensional reconstruction is completed.

Images