An automatic calibration method for dual laser scanning system

Abstract

The present invention discloses an automatic calibration method for a dual-line laser scanning system, comprising: utilizing two single-line laser scanning modules of the dual-line laser scanning system to scan a calibration block respectively to obtain two 3D point cloud data; for the two 3D point cloud data, respectively calculating the normal vector corresponding to the upper plane of each polygonal pyramid to calculate the rotation matrix; respectively projecting the two 3D point cloud data onto the XOY plane, and converting the Z-axis coordinate corresponding to each point into a grayscale value with the X-axis and Y-axis coordinates of the point as the pixel point to establish a 2D image distinguished by grayscale values; utilizing the Harris algorithm to respectively extract the corner points corresponding to the vertices of the upper plane of each polygonal pyramid from the 2D image, then calculating the 3D coordinates corresponding to each corner point, calculating the translation matrix corresponding to the same corner point, and taking the average value as the final translation matrix; utilizing the rotation matrix and the translation matrix to calibrate the 3D point cloud data A and B.

Description

Technical Field

[0001] The present invention belongs to the technical field of machine vision, and in particular relates to an automatic calibration method for a dual-line laser scanning system. Background Art

[0002] With the integration of electronics and optical technologies, non-contact optical measurement methods are becoming increasingly common. Among them, laser line scanning technology is widely used in the field of parts inspection due to its advantages of non-contact, high speed, and high precision.

[0003] A line laser sensor typically consists of two components: a laser that emits a line laser and a photosensitive camera that receives the reflected laser light. The laser beam emitted by the laser is focused by a reflective lens and then incident on the surface of the object being measured. The reflected light is then imaged by a receiving lens onto a photosensitive chip within the camera. The depth of the object being measured is calculated based on the different photosensitive areas of the chip. This measurement method is also known as triangulation due to the angle between the laser and the receiving lens.

[0004] Although single-line laser scanning can meet most inspection requirements in practice, due to the inherent flaws of the triangulation principle, when inspecting workpieces with stepped structures or complex workpieces with upper and lower layered features, the reflected light path is often blocked by the workpiece surface, resulting in the inability of the photosensitive camera to receive the reflected laser information, causing the loss of point cloud data. This leads to incomplete reconstruction of the overall point cloud features, thus affecting feature extraction and measurement accuracy, and failing to meet the requirements of high-precision inspection of workpieces with complex surfaces. Currently, dual-line laser calibration is more often performed manually using manual point selection, which is complex and time-consuming. Summary of the Invention

[0005] The present invention provides an automatic calibration method for a dual-line laser scanning system. With the help of a calibration block with multiple polygonal prisms, the automatic calibration of two single-line laser camera coordinate systems is realized. The calibration process is simple, time-saving, and highly accurate. As a result, the dual-line laser scanning system can scan and detect workpieces with stepped structures and complex workpieces with upper and lower layered features, such as grid components, thereby improving the measurement accuracy of the detection equipment.

[0006] The present invention can be achieved through the following technical solutions:

[0007] An automatic calibration method for a dual-line laser scanning system comprises the following steps:

[0008] Step 1: Collect point cloud data

[0009] The calibration block is placed on the workbench of the dual-line laser scanning system, and the calibration block is scanned by two single-line laser scanning modules of the dual-line laser scanning system to obtain two three-dimensional point cloud data, which are respectively recorded as A and B. The calibration block includes multiple polygonal pyramids on the same plane, and each polygonal pyramid has the same shape;

[0010] Step 2: Calculate the rotation matrix R

[0011] For the three-dimensional point cloud data A and B, the normal vector corresponding to the upper plane of each polygon is calculated and recorded as I m (X m ,Y m ,Z m ), I n (X n ,Y n ,Z n ), where m, n = 1, 2…N, N represents the number of polygons, and the rotation matrix R is calculated using the following equation;

[0012]

[0013] Step 3: Calculate the translation matrix T

[0014] The three-dimensional point cloud data A and B are vertically projected onto the XOY plane, and the Z-axis coordinate corresponding to each point is converted into the grayscale value of the pixel with the X and Y axis coordinates of the point as the pixel. The two-dimensional images A' and B' with grayscale values ​​are established. The Harris feature point extraction algorithm is used to extract the corner points corresponding to the vertices of each polygonal plane from the two-dimensional images A' and B', and then the three-dimensional coordinates corresponding to each corner point are calculated, which are recorded as (X i ,Y i ,Z i )、(X w ,Y w ,Z w ), where i, w = 1, 2, 3...K, K represents the number of vertices of all polygonal pyramids. Then use the following equation to calculate the translation matrix corresponding to the same corner point, and take the average value as the final translation matrix T;

[0015]

[0016]

[0017] Step 4: Use the rotation matrix R and translation matrix T to calibrate the three-dimensional point cloud data A and B.

[0018] Furthermore, the calibration block includes a flat plate, on which a plurality of polygonal pyramids are evenly spaced, and the number of corners of each polygonal pyramid is greater than 3.

[0019] Furthermore, the transmitting ends of the two single-line laser scanning modules are arranged opposite to each other, and the receiving ends are arranged facing each other.

[0020] Furthermore, according to the height of the polygonal pyramid, the three-dimensional point cloud data corresponding to the upper plane of each polygonal pyramid is extracted, and then the normal vector of the plane where the upper plane is located is calculated.

[0021] Furthermore, the X-axis coordinate and the Y-axis coordinate of the corner point are obtained according to the pixel coordinate information of the two-dimensional image, and the Z-axis coordinate is obtained according to the height difference between the flat plate of the calibration block and the upper plane of the polygonal platform.

[0022] Furthermore, the dual-line laser scanning system includes a workbench arranged on a first slide, the first slide is arranged on a first guide rail, the first guide rail is arranged on a vibration isolation platform and arranged along the Y-axis direction, a portal column is also provided on the vibration isolation platform, a second guide rail is provided on the crossbeam of the portal column, the second guide rail is arranged along the X-axis direction, a second slide is arranged on it, the second slide is connected to a third guide rail, the third guide rail is arranged along the Z-axis direction, a third slide is arranged on it, the third slide is connected to the dual-line laser scanning module, and the emitting end of the dual-line laser scanning module is arranged toward the workbench.

[0023] The beneficial technical effects of the present invention are as follows:

[0024] The automatic calibration method of the present invention can realize the rapid calibration and unification of the camera coordinate systems of the two line lasers. In the field of high-precision detection, especially for workpieces with stepped structures and complex workpieces with upper and lower layered features, it can realize fast, efficient and high-precision detection, improve the integrity of data collection, and ultimately ensure the accuracy and efficiency of the detection operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0026] Figure 2 A schematic diagram of the dual-line laser scanning module of the present invention scanning a calibration;

[0027] Figure 3 Schematic diagram of scanning a step-shaped blind spot by a single-line laser scanning module and a dual-line laser scanning module of the present invention, wherein (a) represents a single-line laser scanning module and (b) represents a dual-line laser scanning module;

[0028] Figure 4 Schematic diagram of the structure of the dual-line laser scanning system of the present invention;

[0029] Figure 5 Schematic diagram of the features of corner point recognition of the present invention. DETAILED DESCRIPTION

[0030] The specific implementation of the present invention is described in detail below with reference to the accompanying drawings and preferred embodiments.

[0031] like Figure 1 As shown in the figure, the present invention proposes an automatic calibration method for a dual-line laser scanning system. With the help of a special calibration block, two line laser scanning modules are used to scan and obtain two 3D point cloud data. Then, the rotation matrix R and translation matrix T are obtained respectively by using normal vector transformation and corner point transformation to complete the calibration of the two 3D point cloud data and realize the calibration of the dual-line laser scanning system. The details are as follows:

[0032] Step 1: Collect point cloud data

[0033] Place the calibration block on the workbench of the dual-line laser scanning system, as shown in the figure. Figure 2 As shown, the calibration block is scanned by two line laser scanning modules of the dual-line laser scanning system to obtain two three-dimensional point cloud data, which are respectively recorded as A and B;

[0034] Considering the computational requirements and production costs, Figure 2 As shown in the figure, the calibration block includes multiple polygonal pyramids on the same plane, which can be set together on the same flat plate. The shape of each polygonal pyramid is the same, and the number of its edges and corners must be greater than three to provide more corner point information for subsequent calculations. For example, the flat plate of the calibration block is 200 mm long and 60 mm wide, and 19 quadrangular pyramids are evenly spaced on it. The height of each quadrangular pyramid is 5 mm, the side length is 10 mm, and the diameter of the circular hole is 2 mm.

[0035] Considering the laser scanning of the object with complex structure, there are often blind spots with different structures, such as step-shaped blind spots. Figure 3 As shown, two single-line laser scanning modules are used for data acquisition, with their transmitting ends set opposite to each other and their receiving ends set facing each other. In this way, the emitted light in one direction can be received from two directions, that is, one position of the object to be measured can receive laser scans in two different directions, thereby increasing the probability of being recognized, reducing the impact of blind spots on laser scanning, and improving recognition accuracy.

[0036] Considering the convenience of testing, such as Figure 4As shown, the dual-line laser scanning system includes a three-degree-of-freedom mobile platform, which is composed of a guide rail and a slider, including a workbench arranged on a first slide, the first slide is arranged on the first guide rail, the first guide rail is arranged on the vibration isolation platform, and is arranged along the Y-axis direction. A portal column is also provided on the vibration isolation platform, and a second guide rail is provided on the crossbeam of the portal column. The second guide rail is arranged along the X-axis direction, and a second slide is arranged thereon. The second slide is connected to the third guide rail, and the third guide rail is arranged along the Z-axis direction, and a third slide is arranged thereon. The third slide is connected to the dual-line laser scanning module, and the transmitting end of the dual-line laser scanning module is arranged toward the workbench, including two single-line laser scanning modules, whose transmitting ends are arranged relative to each other, and whose receiving ends are arranged facing each other. In this way, with the help of a three-degree-of-freedom mobile platform, adjustments in three directions of the object to be measured and the dual laser scanning module can be completed, so that the measurement of the object to be measured can be carried out in all directions. With the help of the dual laser scanning module, scanning of blind spots in some structures such as steps can be realized, thereby completing the measurement of workpieces with complex structures such as gate components, providing an accurate data basis for subsequent calculations.

[0037] Step 2: Calculate the rotation matrix R

[0038] For the three-dimensional point cloud data A and B, the normal vector corresponding to the upper plane of each polygon is calculated and recorded as I m (X m ,Y m ,Z m ), I n (X n ,Y n ,Z n ), where m, n = 1, 2…N, N represents the number of polygons, and the rotation matrix R is calculated using the following equation;

[0039]

[0040] in, is the transfer matrix between two normal vectors. Since the transfer matrix between normal vectors only represents the transformation between two normal vectors and has no reference to the results, it is not considered.

[0041] Since the two point cloud data obtained contain point cloud noise, manual denoising or Gaussian filtering is required to simplify the point cloud data. Since the point cloud data is in the form of three-dimensional coordinate values ​​(x, y, z), we can set the bottom surface of the calibration block on the XOY plane. Then, using the height information in the Z axis direction as a distinction, after filtering the Z height information with a 5.000mm threshold, we can extract the three-dimensional point cloud data where the upper plane of the pyramid is located, and then calculate the normal vectors of each upper plane.

[0042] Step 3: Calculate the translation matrix T

[0043] The three-dimensional point cloud data A and B are vertically projected onto the XOY plane, and the Z-axis coordinates corresponding to each point are converted into the grayscale value of the pixel with the X and Y axis coordinates of the point, and the two-dimensional images A' and B' with grayscale values ​​are established. The PCL algorithm can be used to calculate the ModelCoefficients of the plane model projection filter algorithm. Then, the Harris feature point extraction algorithm is used to extract the corner points corresponding to the vertices of each polygonal plane from the two-dimensional images A' and B', and then the three-dimensional coordinates corresponding to each corner point are calculated, which are recorded as (X i ,Y i ,Z i )、(X w ,Y w ,Z w ), where i, w = 1, 2, 3...K, K represents the number of vertices of all polygonal pyramids. Then use the following equation to calculate the translation matrix corresponding to the same corner point, and take the average value as the final translation matrix T;

[0044]

[0045]

[0046] From the features of the calibration block, we know that the upper plane of each polygon has four corner points, with a total of 76 corner points. These corner points are marked as 1-76 in sequence. The Harris feature point extraction algorithm is used to obtain the 76 corner point information including position information. The corner point position can be determined according to the gradient change of the image grayscale value in the x and y directions. The following formula is used to determine whether it is a corner point:

[0047]

[0048]

[0049] Among them, E(u,v) represents the change of gray value with the change of coordinate offset (u,v). When the coordinate (u,v) changes, the larger E(u,v) is, the better the distinction is. M is represented as a window function. By statistically analyzing the gradient of each pixel in the window in the x direction and the gradient in the y direction, the gradient coordinate of each pixel point is expressed as (I x ,I yDefine λ1 as the gradient change in the x direction, λ2 as the gradient change in the y direction, and set a certain threshold, here the threshold is set to ω=0.5, when the gradient changes of λ1 and λ2 are λ1<ω,λ2<ω, it is judged as a flat area; when the gradient changes of λ1 and λ2 are λ1>ω,λ2>ω, it is judged as a corner point; when only one direction of the gradient changes to λ1>ω or λ2>ω, it is judged as an edge, such as Figure 5 shown.

[0050] Since the corner point calculation is based on a two-dimensional image projected from a three-dimensional point cloud plane, the Z-axis height information of the corner point is obtained based on the height difference between the flat plate of the calibration block and the upper plane of the polygonal table, which is 5.000 mm. The three-dimensional coordinates of each corner point in the two-dimensional image can be obtained.

[0051] Step 4: Use the rotation matrix R and translation matrix T to calibrate the three-dimensional point cloud data A and B, thereby realizing the calibration of the dual-line laser scanning system.

[0052] Although specific embodiments of the present invention are described above, those skilled in the art should understand that these are merely examples and that various changes or modifications may be made to these embodiments without departing from the principles and essence of the present invention. Therefore, the scope of protection of the present invention is limited by the appended claims.

Claims

1. An automatic calibration method for a dual-line laser scanning system, characterized in that The following steps are involved: Step 1: Collect point cloud data The calibration block is placed on the workbench of the dual-line laser scanning system, and the calibration block is scanned by two single-line laser scanning modules of the dual-line laser scanning system to obtain two three-dimensional point cloud data, which are respectively recorded as A and B. The calibration block includes multiple polygonal pyramids on the same plane, and each polygonal pyramid has the same shape; Step 2: Calculate the rotation matrix R For the three-dimensional point cloud data A and B, the normal vector corresponding to the upper plane of each polygon is calculated and recorded as 、 , where m=1,2…N, n=1,2…N, N represents the number of polygons. Use the following equation to calculate the rotation matrix R; ， ， in, is the transfer matrix between the two normal vectors from the 3D point cloud data A and B; Step 3: Calculate the translation matrix T The three-dimensional point cloud data A and B are vertically projected onto the XOY plane, and the Z-axis coordinates corresponding to each point are converted into the grayscale value of the pixel with the X and Y axis coordinates of the point as the pixel, and the two-dimensional images A' and B' with grayscale values ​​are established. Harris The feature point extraction algorithm extracts the corner points corresponding to the vertices of each polygonal plane from the two-dimensional images A' and B', and then calculates the three-dimensional coordinates corresponding to each corner point, which are recorded as 、 , where i=1,2,3…K, w=1,2,3…K, K represents the number of vertices of all polygonal pyramids, and then use the following equation to calculate the translation matrix corresponding to the same corner point, and take the average value as the final translation matrix T; Step 4: Use the rotation matrix R and translation matrix T to calibrate the three-dimensional point cloud data A and B to achieve the calibration of the dual-line laser scanning system.

2. The automatic calibration method for a dual-line laser scanning system according to claim 1, characterized in that: The calibration block includes a flat plate, on which a plurality of polygonal pyramids are evenly spaced, and the number of corners of each polygonal pyramid is greater than 3.

3. The automatic calibration method for a dual-line laser scanning system according to claim 1, characterized in that: The transmitting ends of the two single-line laser scanning modules are arranged opposite to each other, and the receiving ends are arranged facing each other.

4. The automatic calibration method for a dual-line laser scanning system according to claim 2, characterized in that: According to the height of the polygonal pyramid, the three-dimensional point cloud data corresponding to the upper plane of each polygonal pyramid is extracted, and then the normal vector of the plane where each upper plane is located is calculated.

5. The automatic calibration method for a dual-line laser scanning system according to claim 2, characterized in that: The X-axis coordinate and the Y-axis coordinate of the corner point are obtained according to the pixel coordinate information of the two-dimensional image, and the Z-axis coordinate is obtained according to the height difference between the flat plate of the calibration block and the upper plane of the polygonal platform.

6. The automatic calibration method for a dual-line laser scanning system according to claim 1, characterized in that: The dual-line laser scanning system includes a workbench arranged on a first slide, the first slide is arranged on a first guide rail, the first guide rail is arranged on a vibration isolation platform and arranged along the Y-axis direction, a portal column is also arranged on the vibration isolation platform, a second guide rail is arranged on the crossbeam of the portal column, the second guide rail is arranged along the X-axis direction, and a second slide is arranged on it, the second slide is connected to a third guide rail, the third guide rail is arranged along the Z-axis direction, and a third slide is arranged on it, the third slide is connected to the dual-line laser scanning module, and the emitting end of the dual-line laser scanning module is arranged toward the workbench.

Images