Non-contact coaxiality measurement method and system

Abstract

The present invention provides a non-contact coaxiality measurement method and system, the system includes a single-axis drive module, an industrial scanning device, a laser emitting device and a data processing module; the system is used to implement the following steps: obtaining structured light images of a measured part and a reference part; inputting the structured light images into an optimized three-dimensional reconstruction model, and extracting point cloud data of the measured part and the reference part using the optimized three-dimensional reconstruction model; according to the point cloud data of the measured part and the reference part, obtaining the cylindrical curve of the measured part and the cylindrical curve of the reference part, and using the common axis method to calculate the coaxiality of the two cylindrical curves as coaxiality measurement data. The laser scanning measurement form is adopted to realize non-contact measurement, simplify the measurement procedure, save measurement time, and the measurement process is simple and easy to implement.

Description

Technical Field

[0001] The present invention relates to the technical field of motor coaxiality detection, and in particular to a non-contact coaxiality measurement method and system. Background Art

[0002] The traditional method of measuring the coaxiality of the inner hole of the motor has always been to use the three-coordinate dot method for measurement. When using the probe, place the product horizontally or vertically, fix the product on the platform, and calibrate it through a standard ball, ring gauge or standard block to determine the diameter and accuracy of the spherical ruby ​​probe. After the probe is calibrated, dot the inner wall of the inner hole (the number of points is confirmed based on the actual situation), and record the three-dimensional coordinates of the points in three-dimensional space at the same time. The measurement software processes the point cloud data to complete the measurement. Contact measurement is usually used, that is, the probe contacts the measured part, and the three-dimensional coordinate value of the measured point is obtained through the displacement sensor of the probe.

[0003] At present, three-dimensional coordinate measuring machines are mostly used for this type of measurement. Three-dimensional coordinate measuring machines have high measurement accuracy and good repeatability. They can easily and accurately evaluate dimensional data and obtain the coordinate positions of each measuring point on the measured object. According to the spatial coordinate values ​​of these points, the geometric size, shape and position of the measured object can be calculated. It can provide real feedback on product information and is recognized as benchmarking data in the measurement industry. However, three-dimensional coordinate measurement also has the following disadvantages, which limit the cost reduction and efficiency improvement of industrial production:

[0004] 1. Due to the different measurement methods and measurement bases, as well as the influence of factors such as sampling strategies and measurement environment, large measurement errors and poor repeatability often occur;

[0005] 2. The detection cycle is slow. Usually, the measurement time of a workpiece ranges from 5 to 20 minutes, which seriously affects production efficiency;

[0006] 3. Improper operation can easily damage the surface accuracy of some important parts of the workpiece and also damage the probe;

[0007] 4. Expensive, difficult to purchase and use in large quantities;

[0008] 5. The environment requirements are stringent, requiring a laboratory environment. It has very high requirements on the temperature, humidity, and levelness of the objects being tested, and cannot be applied to industrial production lines. Summary of the invention

[0009] Therefore, the purpose of the present invention is to provide a non-contact coaxiality measurement method and system, which adopts the form of laser scanning detection to realize non-contact measurement, simplify the measurement procedure, save measurement time, and the measurement process is simple and easy to implement; the coaxiality measurement method of the present invention is not only suitable for the measurement of motor coaxiality, but also suitable for other fields involving coaxiality measurement.

[0010] To this end, the present invention provides a non-contact coaxiality measurement method, comprising:

[0011] S1, obtaining structured light images of the test piece and the reference piece;

[0012] S2, inputting the structured light image into the optimized three-dimensional reconstruction model, and using the optimized three-dimensional reconstruction model to extract point cloud data of the measured object and the reference object; the establishment and optimization of the three-dimensional reconstruction model includes the following process:

[0013] S201, calibrating the camera's internal parameters and light plane to form an initial 3D reconstruction model;

[0014] S202, calibrating the translation stage, and compensating the translation stage calibration result to the initial three-dimensional reconstruction model to eliminate installation errors;

[0015] S203, establishing a coaxiality residual function, optimizing the calibrated light plane parameters and the three-dimensional reconstruction model after eliminating the installation error, and obtaining an optimized three-dimensional reconstruction model;

[0016] S3. According to the point cloud data of the measured object and the reference object, the cylindrical curve of the measured object and the cylindrical curve of the reference object are obtained, and the coaxiality of the two cylindrical curves is calculated as the coaxiality measurement data.

[0017] Further preferably, in S201, calibrating the intrinsic parameters and light plane of the camera includes:

[0018] Obtain the calibration image required for light plane calibration, calibrate the light plane using the point-plane intersection method, and construct the spatial plane equation of the light plane;

[0019] Performing initial light plane fitting according to the calibration image and the constructed space plane equation to obtain a first fitting light plane;

[0020] Measuring the distance from the three-dimensional point of the light strip to the first fitting light plane;

[0021] According to the preset error threshold, the abnormal points in the three-dimensional points of the light strip whose distance from the first fitting light plane exceeds the preset error threshold are filtered out, and the retained data points are used to perform secondary light plane fitting to obtain the final calibrated light plane parameters.

[0022] Further preferably, in S202, calibrating the translation stage comprises the following steps:

[0023] S2021. Place a calibration plate in the camera field of view, keep the calibration plate stationary, and move the translation stage so that the camera sequentially takes multiple images of the calibration plate driven by the translation stage;

[0024] S2022, extracting characteristic corner points of each calibration plate image, obtaining external parameters corresponding to the calibration plate coordinate system to the camera coordinate system, and calculating the corresponding camera optical center position parameters when obtaining each calibration plate image;

[0025] S2023, converting the optical center position parameters of the multiple cameras into the calibration plate coordinate system, eliminating errors, and performing linear fitting on the optical center position parameters of the multiple cameras by using the least squares method;

[0026] S2024. Calculate the direction vector of the fitting line, and normalize the obtained direction vector to obtain the calibration result of the translation stage.

[0027] Further preferably, in S203, a coaxiality residual function is established to optimize the calibrated light plane parameters and the three-dimensional reconstruction model after eliminating the installation error, including the following process:

[0028] S2031, placing a standard part with known coaxiality within the camera field of view, adjusting the standard part at different positions, scanning and three-dimensionally reconstructing the standard part at each position;

[0029] S2032, optimizing the target parameters using the established coaxiality residual function, and when the residual function value is minimum, the target parameters obtain the optimal solution;

[0030]

[0031] Among them, H is the camera internal parameter matrix, k1, k2, k3, p1, p2 are camera distortion coefficients, A, B, D are light plane parameters, k x is the projection coefficient of the angle between the camera axis and the stage, k y is the systematic error; y i The coaxiality of standard parts in different placement positions.

[0032] Further preferably, in S3, it also includes fitting the cylindrical curves of the measured part and the reference part using the least squares method according to the extracted point cloud data of the measured part and the reference part.

[0033] Further preferably, in S3, the coaxiality of the two cylindrical curves is calculated based on the cylindrical curve of the measured object and the cylindrical curve of the reference object obtained by fitting, and the coaxiality measurement data includes:

[0034] In the fitted cylindrical curve of the measured object and the cylindrical curve of the reference object, the upper, middle and lower endpoints of the central axis of the cylindrical curve of the measured object are obtained, which are recorded as points A, B and C; and the upper, middle and lower endpoints of the central axis of the cylindrical curve of the reference object are obtained, which are recorded as points D, E and F;

[0035] According to the obtained six points A, B, C, D, E, and F, the common axis is fitted, the point farthest from the common axis is selected, and the vertical distance d from the farthest point to the common axis is calculated. The coaxiality is 2d.

[0036] The present invention also provides a non-contact coaxiality measurement system for implementing the above-mentioned non-contact coaxiality measurement method, comprising a single-axis drive module, an industrial scanning device, a laser emitting device and a data processing module;

[0037] The single-axis driving module is provided with a translation stage for installing the industrial scanning equipment and driving the industrial scanning equipment and the laser emitting device to gradually move forward or backward to scan the measured object and the reference object;

[0038] The industrial scanning device is used to scan the measured object and the reference object, and obtain the scanned structured light image in real time;

[0039] The laser emitting device comprises a point laser and a total reflection cone mirror, and the laser emitting device is located directly below the industrial scanning equipment and is coaxially installed with the camera lens;

[0040] The data processing module is used to extract the point cloud data of the measured object and the reference object based on the scanned structured light image and the optimized three-dimensional reconstruction model; obtain the cylindrical curve of the measured object and the cylindrical curve of the reference object, and use the common axis method to calculate the coaxiality of the two cylindrical curves as the final coaxiality.

[0041] Further preferably, in the laser emitting device, the point laser is arranged on the top of the total reflective conical mirror, and when the point laser emits laser, the laser spot is reflected by the total reflective conical mirror to form a circular ring structured light.

[0042] Compared with the prior art, the non-contact coaxiality measurement method and system disclosed in the present application have at least the following advantages:

[0043] 1. This application uses a point laser and a conical semi-reflective and semi-transparent cone mirror to form a circular ring laser. When the inner wall contour of the measured object blocks the laser ring, a circular image is formed. The formed circular image is calibrated using the established light plane. When the single-axis drive module scans the inner wall of the measured object up and down, the calibrated circular images are gradually superimposed to complete the construction of the three-dimensional reconstruction model. This form of laser scanning detection realizes non-contact measurement, simplifies the measurement procedure, saves measurement time, and the measurement process is simple and easy to implement.

[0044] 2. By calibrating the translation stage, the calibration result of the translation stage is compensated to the 3D reconstruction model to eliminate the installation error; the light signal change caused by the modulation of the line structured light by the object contour is reconstructed through accurate and efficient contour extraction technology. After the 3D reconstruction of the product shape is completed, when measuring the coaxiality, the coaxiality of the measured part is calculated using the coaxiality of the standard part to obtain the relevant parameters of the light plane and the camera, and then the point cloud data is used to calculate the coaxiality of the measured part, with small calculation amount and fast detection speed.

[0045] 3. High precision: The measurement point cloud covers a larger range, reflects the product size information more realistically, avoids misjudgment caused by tiny defects, and the dynamic measurement accuracy reaches the micron level.

[0046] 4. High efficiency: Automated and rapid testing, the measurement of a single product is within 60 seconds, ensuring the rhythm of rapid industrial production.

[0047] 5. Light stripe extraction algorithm: Through the changes in light signals caused by the modulation of line structured light by the object contour, the object contour is reconstructed through precise and efficient contour extraction technology, and the push scan is completed to achieve three-dimensional reconstruction of the entire product shape. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flow chart of a non-contact coaxiality measurement method provided by the present invention;

[0049] Figure 2 Provides a result diagram of abnormal point detection during light plane fitting for the present invention;

[0050] Figure 3 It is a schematic diagram of the calibration result of the displacement stage in the present invention;

[0051] Figure 4 A schematic diagram of three-dimensional point cloud data for coaxiality optimization using standard parts in the present invention;

[0052] Figure 5 A schematic diagram of three-dimensional point cloud data for coaxiality optimization using standard parts from another angle;

[0053] Figure 6 This is the error curve before the coaxiality residual optimization function is optimized;

[0054] Figure 7 It is the error curve after the coaxiality residual optimization function is optimized;

[0055] Figure 8 Schematic diagram of calculating the coaxiality of two cylindrical curves using the common axis method;

[0056] Fig. 9 It is a structural schematic diagram of the non-contact coaxiality measurement system of the present invention;

[0057] Fig.10 This is a schematic diagram of the imaging principle of the industrial laser of the present invention;

[0058] In the figure: 1-single-axis drive module, 2-industrial camera, 3-industrial lens, 4-laser emitting device, 401-point laser, 402-total reflection conical mirror, 5-sample to be tested. DETAILED DESCRIPTION

[0059] The present invention is further described in detail below through the accompanying drawings and specific embodiments.

[0060] In this embodiment, based on the principle of optical non-contact measurement, an optoelectronic information acquisition system based on point laser, cone mirror, module and camera is built, and the installation error is eliminated in the most optimized way. The circular structure light system is calibrated by a new calibration method. This system uses a point laser and a conical semi-reflective semi-transparent cone mirror to form a circular ring laser. When the inner wall contour of the measured object blocks the laser ring, a circular image is formed. The formed circular image is calibrated using the established light plane. When the single-axis drive module scans the inner wall of the measured object up and down, the calibrated circular image is gradually superimposed to complete the construction of the three-dimensional reconstruction model. By calibrating the displacement stage, the calibration result of the displacement stage is compensated to the three-dimensional reconstruction model, eliminating the system error; through the change of the light signal caused by the modulation of the line structure light by the object contour, the object contour is reconstructed through accurate and efficient contour extraction technology, and the push scanning is completed to achieve the three-dimensional reconstruction of the entire product morphology, and then the coaxiality of the sample is measured. After the coaxiality of the standard part is calculated to obtain the relevant parameters of the light plane and the camera, the point cloud data can be used to calculate the coaxiality of the measured part, with small calculation amount and fast detection speed.

[0061] Embodiment 1, as Figure 1 As shown, an embodiment of the present invention provides a non-contact coaxiality measurement method, comprising:

[0062] S1, obtaining structured light images of the test piece and the reference piece;

[0063] S2, inputting the structured light image into the optimized three-dimensional reconstruction model, and using the optimized three-dimensional reconstruction model to extract point cloud data of the measured object and the reference object;

[0064] The establishment and optimization of the three-dimensional reconstruction model includes the following processes:

[0065] S201, calibrating the camera's internal parameters and light plane, specifically including:

[0066] Obtain the calibration image required for light plane calibration, calibrate the light plane using the point-plane intersection method, and construct the spatial plane equation of the light plane;

[0067] Performing initial light plane fitting according to the calibration image and the constructed space plane equation to obtain a first fitting light plane;

[0068] Measure the distance from the light strip 3D point to the first fitting light plane; it should be noted that the light strip 3D point is a set of specific points in space formed by the laser lines projected by the line structured light projector onto the object to be measured when using a line structured light 3D scanning system. The coordinates of these points in 3D space can be calculated by the corresponding imaging system and algorithm.

[0069] Specifically, the line structured light 3D scanning system consists of a camera and a line structured light projector. The line structured light projector projects the laser onto the object to be measured in a linear pattern to form a bright light strip. The light strip deforms as the geometric shape of the object surface changes. The camera captures the deformed light strip image from a certain angle. By analyzing the deformation of the light strip on the camera imaging plane and known system parameters (such as the focal length of the camera, the relative position of the light strip and the camera, etc.), the three-dimensional coordinates of the corresponding points on the object surface can be calculated. Therefore, the abnormal points can be judged based on the distance from the three-dimensional point of the light strip to the first fitting light plane. The present application reconstructs the object contour through the changes in light signals caused by the modulation of the line structured light by the object contour through precise and efficient contour extraction technology, and the push scan is completed to achieve three-dimensional reconstruction of the entire product morphology.

[0070] According to the preset error threshold, the abnormal points in the three-dimensional points of the light strip whose distance from the first fitting light plane exceeds the preset error threshold are filtered out, and the retained data points are used for secondary light plane fitting to obtain the final calibrated light plane parameters. In light plane fitting, the accuracy of the least squares algorithm is easily affected by isolated points introduced by scratches, stains, vibrations and other factors on the inner wall of the workpiece. Directly fitting the plane with all data containing gross errors will lead to low reliability of the results. Therefore, according to the preset error threshold, the 3σ criterion is used to eliminate gross errors. When fitting the plane, when the distance from a measurement point to the fitted plane is greater than 3σ, it is an abnormal point; such as Figure 2 As shown, it is a schematic diagram of the result of light plane fitting.

[0071] Specifically, it includes obtaining the calibration image required for light plane calibration. Before setting up the system, the camera needs to be calibrated internally. Both MATLAB and OpenCV can be used to calibrate the conventional camera internal parameters f x 、f y , u0, v0, k1, k2, k3, p1, p2;

[0072]

[0073] The point-plane intersection method is used to construct the spatial plane equation of the light plane and calibrate the light plane;

[0074]

[0075] After calibration, the initial three-dimensional reconstruction model is obtained as shown in the following formula (3):

[0076]

[0077] Among them, Xc, Yc, and Zc are the coordinate values ​​of the XYZ axes of the three-dimensional point cloud respectively; f x 、f y , u0, v0, are the internal parameters of the camera; k1, k2, k3, p1, p2 are the distortion coefficients of the camera; A, B, C, D are the constants in the spatial plane equation of the established light plane;

[0078] Furthermore, in order to reduce the installation errors caused by the structural and principle defects of the measuring instrument and the external environment, it is necessary to calibrate the translation stage to ensure the accuracy and reliability of the measurement results. Therefore, the following operations are included:

[0079] S202, calibrate the translation stage, and compensate the translation stage calibration result to the initial 3D reconstruction model to eliminate the installation error; Figure 3 As shown, specifically including:

[0080] S2021. Place a calibration plate in the camera field of view, keep the calibration plate stationary, move the translation stage, and allow the camera to take multiple pictures of the calibration plate in sequence driven by the translation stage; wherein the calibration plate adopts a checkerboard, and the multiple calibration plate images in the present application are pictures of the checkerboard, and the camera is used to take pictures of the checkerboard driven by the translation stage, and multiple checkerboard images of different sizes are obtained due to the different distances between the camera and the checkerboard.

[0081] S2022, extracting characteristic corner points of each calibration plate image, obtaining external parameters corresponding to the calibration plate coordinate system to the camera coordinate system, and calculating the corresponding camera optical center position parameters when obtaining each calibration plate image;

[0082] S2023, converting the optical center position parameters of the multiple cameras into the calibration plate coordinate system, eliminating errors, and performing linear fitting on the optical center position parameters of the multiple cameras by using the least squares method;

[0083] S2024. Calculate the direction vector of the fitting line, and normalize the obtained direction vector to obtain the calibration result of the translation stage.

[0084] Due to factors such as installation, it is impossible to ensure that the displacement platform moves strictly in the direction of the camera optical axis, so the displacement stage calibration is required. Place the calibration plate on the loading platform, and move the displacement stage so that the calibration plate is within the camera depth of field. According to the actual working direction, take several calibration plate pictures in sequence. Extract the characteristic corner points of each calibration plate picture, use the slovePnP method to obtain the corresponding external parameters from the calibration plate coordinate system to the camera coordinate system, place the calibration plate unused, and transform the camera optical center pose at multiple positions to the calibration plate coordinate system. After eliminating the gross errors of multiple optical center points, fit the straight line through the least squares method, and obtain the direction vector of the straight line and normalize it to obtain the displacement stage calibration result and direction vector. Compensate the direction vector to the three-dimensional reconstructed point cloud to preliminarily eliminate the installation error. That is, the deviation distance and direction of the centers of the two circles in the figure are the installation errors.

[0085] like Figure 4-Figure 7 As shown, S203, establishing a coaxiality residual optimization function, optimizing the calibrated light plane parameters and the three-dimensional reconstruction model after eliminating the installation error, to obtain an optimized three-dimensional reconstruction model; specifically including the following processes:

[0086] S2031, placing a standard part with known coaxiality within the camera field of view, adjusting the standard part at different positions, scanning the standard part at each position and performing three-dimensional reconstruction; Figure 4 and Figure 5 As shown, a standard part with known coaxiality is placed within the camera field of view, and the standard part is adjusted to different positions, and the positions need to cover all positions within the camera field of view. For example, the standard part is placed at 49 positions within the camera field of view, and the standard part at each position is scanned and three-dimensionally reconstructed for subsequent installation error optimization;

[0087] S2032. Optimize the target parameters using the established coaxiality residual optimization function, and calculate the minimum residual between the predicted value and the actual value; when the residual function value is the minimum, the target parameters obtain the optimal solution.

[0088]

[0089] Among them, H is the camera internal parameter matrix, k1, k2, k3, p1, p2 are camera distortion coefficients, A, B, D are light plane parameters, k x is the projection coefficient of the angle between the camera axis and the stage, k y is the system error; yi is the coaxiality of the standard parts at different placement positions; where k1, k2, k3, p1, p2, H, k x , k y These are target parameters.

[0090] Using the coaxiality residual function as the objective function, the calibrated light plane parameters and the 3D reconstruction model after eliminating the installation error are optimized. The optimization results are shown in Figure 2. Figure 6 and Figure 7 shown.

[0091] S3. According to the point cloud data of the measured object and the reference object, the cylindrical curve of the measured object and the cylindrical curve of the reference object are obtained, and the coaxiality of the two cylindrical curves is calculated using the common axis method as the coaxiality measurement data. Figure 8 shown.

[0092] Specifically, it includes: according to the extracted point cloud data of the measured object and the reference object, the cylindrical curves of the measured object and the reference object are obtained by fitting the least square method;

[0093] According to the fitting, the cylindrical curve of the measured part and the cylindrical curve of the reference part are obtained, and the coaxiality of the two cylindrical curves is calculated by the common axis method. The coaxiality measurement data includes:

[0094] In the fitted cylindrical curve of the measured object and the cylindrical curve of the reference object, the upper, middle and lower endpoints of the central axis of the cylindrical curve of the measured object are obtained, which are recorded as points A, B and C; and the upper, middle and lower endpoints of the central axis of the cylindrical curve of the reference object are obtained, which are recorded as points D, E and F;

[0095] According to the obtained six points A, B, C, D, E, and F, a common axis is fitted, the point farthest from the common axis is selected, and the vertical distance d from the farthest point to the common axis is calculated. The coaxiality is 2d.

[0096] like Figure 9-10 As shown, the present invention also provides a non-contact coaxiality measurement system for implementing the above-mentioned non-contact coaxiality measurement method, comprising a single-axis drive module 1, an industrial scanning device, a laser emitting device 4 and a data processing module; wherein the industrial scanning device comprises an industrial camera 2 and an industrial lens 3; the laser emitting device 4 comprises a point laser 401 and a conical total reflection cone mirror 402;

[0097] The single-axis driving module 1 is provided with a translation stage for installing the industrial scanning equipment and driving the industrial scanning equipment and the laser emitting device 4 to gradually move forward or backward to scan and detect the measured object and the reference object;

[0098] The industrial scanning device is used to scan the measured object and the reference object, and obtain the scanned structured light image in real time;

[0099] like Fig.10The laser emitting device 4 includes a point laser 401 and a total reflection cone mirror 402. The laser emitting device 4 is located directly below the industrial scanning equipment. The point laser 401 is arranged on the top of the total reflection cone mirror 402. The total reflection cone mirror is conical. When the point laser 401 emits laser, the laser spot is reflected by the conical total reflection cone mirror to form a circular structure light.

[0100] The data processing module is used to extract the point cloud data of the measured object and the reference object based on the scanned structured light image and the optimized three-dimensional reconstruction model; obtain the cylindrical curve of the measured object and the cylindrical curve of the reference object, and use the common axis method to calculate the coaxiality of the two cylindrical curves as the final coaxiality.

[0101] The specific implementation process of the system refers to the above method embodiment, which will not be repeated here.

[0102] The non-contact coaxiality detection adopted in this application uses optical or laser technology to evaluate the coaxiality of the target object by measuring the surface features or position of the target object. This method avoids direct contact with the target object, thereby reducing interference with the object being measured and achieving high-precision detection. This technology can be applied to the coaxiality detection of tram motors. The laser source hits the inner wall of the motor vertically to form a ring light. The camera located on the central axis of light diffusion collects the ring image and reconstructs the inner wall profile in three dimensions. The servo module or robotic arm drives the measuring probe to perform scanning motion and reconstructs the entire inner wall profile in three dimensions. Measure the axis position of the rotor or other rotating parts. Finally, the cylindrical feature extraction and optimization fitting algorithm are studied to measure the coaxiality; these devices can quickly and accurately obtain the surface features or position data of the target object, and evaluate the coaxiality through computational analysis. Compared with traditional contact measurement methods, non-contact methods are generally more suitable for measuring high-speed rotating parts and can provide more comprehensive information. The coaxiality of the rotating parts inside the tram motor can be more effectively evaluated, and potential quality problems can be discovered and solved in a timely manner, thereby ensuring that the performance and reliability of the motor meet the design requirements. This advanced testing method helps improve production efficiency and ensures that product quality meets standards.

[0103] Obviously, the above embodiments are only examples for clear explanation, and are not intended to limit the implementation methods. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from them are still within the protection scope of the invention.

Claims

1. A non-contact coaxiality measurement method, characterized in that: include: S1, obtaining structured light images of the test piece and the reference piece; S2, inputting the structured light image into the optimized three-dimensional reconstruction model, and using the optimized three-dimensional reconstruction model to extract point cloud data of the measured object and the reference object; The establishment and optimization of the three-dimensional reconstruction model includes the following processes: S201, calibrating the camera's internal parameters and light plane to form an initial 3D reconstruction model; S202, calibrating the translation stage, and compensating the translation stage calibration result to the initial three-dimensional reconstruction model to eliminate installation errors; S203, establishing a coaxiality residual function, including: S2031, placing a standard part with known coaxiality within the camera field of view, adjusting the standard part at different positions, scanning and three-dimensionally reconstructing the standard part at each position; S2032, optimizing the target parameters using the established coaxiality residual function, and when the residual function value is minimum, the target parameters obtain the optimal solution; RES= Among them, H is the camera intrinsic parameter matrix, is the camera distortion coefficient, is the light plane parameter, is the projection coefficient of the angle between the camera principal axis and the translation stage, is the systematic error; y i Coaxiality of standard parts at different placement positions; Optimizing the calibrated light plane parameters and the three-dimensional reconstruction model after eliminating installation errors to obtain an optimized three-dimensional reconstruction model; S3. According to the point cloud data of the measured object and the reference object, the cylindrical curve of the measured object and the cylindrical curve of the reference object are obtained, and the coaxiality of the two cylindrical curves is calculated as the coaxiality measurement data.

2. The non-contact coaxiality measurement method according to claim 1, characterized in that: In S201, calibrating the intrinsic parameters and light plane of the camera includes: Obtain the calibration image required for light plane calibration, calibrate the light plane using the point-plane intersection method, and construct the spatial plane equation of the light plane; Performing initial light plane fitting according to the calibration image and the constructed space plane equation to obtain a first fitting light plane; Measuring the distance from the three-dimensional point of the light strip to the first fitting light plane; According to the preset error threshold, the abnormal points in the three-dimensional points of the light strip whose distance from the first fitting light plane exceeds the preset error threshold are filtered out, and the retained data points are used to perform secondary light plane fitting to obtain the final calibrated light plane parameters.

3. The non-contact coaxiality measurement method according to claim 1, characterized in that: In S202, the translation stage is calibrated, including the following steps: S2021. Place a calibration plate in the camera field of view, keep the calibration plate stationary, and move the translation stage so that the camera sequentially takes multiple images of the calibration plate driven by the translation stage; S2022, extracting characteristic corner points of each calibration plate image, obtaining external parameters corresponding to the calibration plate coordinate system to the camera coordinate system, and calculating the corresponding camera optical center position parameters when obtaining each calibration plate image; S2023, converting the optical center position parameters of the multiple cameras into the calibration plate coordinate system, eliminating errors, and performing linear fitting on the optical center position parameters of the multiple cameras by using the least squares method; S2024. Calculate the direction vector of the fitting line, and normalize the obtained direction vector to obtain the calibration result of the translation stage.

4. The non-contact coaxiality measurement method according to claim 1, characterized in that: In S3, it also includes fitting the cylindrical curves of the measured object and the reference object using the least square method according to the extracted point cloud data of the measured object and the reference object.

5. The non-contact coaxiality measurement method according to claim 1, characterized in that: In S3, the cylindrical curve of the measured object and the cylindrical curve of the reference object are obtained by fitting, and the coaxiality of the two cylindrical curves is calculated, and the coaxiality measurement data includes: In the fitted cylindrical curve of the measured object and the cylindrical curve of the reference object, the upper, middle and lower endpoints of the central axis of the cylindrical curve of the measured object are obtained, which are recorded as points A, B and C; and the upper, middle and lower endpoints of the central axis of the cylindrical curve of the reference object are obtained, which are recorded as points D, E and F; According to the obtained six points A, B, C, D, E, and F, the common axis is fitted, the point farthest from the common axis is selected, and the vertical distance d from the farthest point to the common axis is calculated. The coaxiality is 2d.

6. A non-contact coaxiality measurement system, characterized in that: A non-contact coaxiality measurement method for implementing any one of claims 1 to 5, comprising a single-axis drive module, an industrial scanning device, a laser emitting device and a data processing module; The single-axis driving module is provided with a translation stage for installing the industrial scanning equipment and driving the industrial scanning equipment and the laser emitting device to gradually move forward or backward to scan the measured object and the reference object; The industrial scanning device is used to scan the measured object and the reference object, and obtain the scanned structured light image in real time; The laser emitting device comprises a point laser and a total reflection cone mirror, and the laser emitting device is located directly below the industrial scanning equipment and is coaxially installed with the camera lens; The data processing module is used to extract the point cloud data of the measured object and the reference object based on the scanned structured light image and the optimized three-dimensional reconstruction model; obtain the cylindrical curve of the measured object and the cylindrical curve of the reference object, and use the common axis method to calculate the coaxiality of the two cylindrical curves as the final coaxiality.

7. The non-contact coaxiality measurement system according to claim 6, characterized in that: In the laser emitting device, the point laser is arranged on the top of the total reflective cone mirror. When the point laser emits laser, the laser spot is reflected by the total reflective cone mirror to form a circular ring structure light.

Images