Steel plate shape measurement and calibration method

Abstract

The invention provides a steel plate shape measurement and calibration method, belongs to the field of three-dimensional shape measurement and calibration, and aims to realize three-dimensional reconstruction and shape measurement of the surface of a steel plate. In order to realize high-precision detection of the shape of the steel plate, X, Y and Z coordinate data of the surface of the steel plate are obtained through a double-line laser measurement system. The measurement system is composed of a plurality of cameras and two line lasers, and the installation positions of the cameras are determined to ensure that a certain overlapping view exists between the cameras, so that data splicing is realized. In the measuring process, an image shot by the camera is processed, position information of laser rays in the image is extracted, pixel coordinates are converted into actual world coordinates according to calibration parameters and the working distance of the camera, and X, Y and Z three-dimensional coordinate data of the surface of the steel plate are obtained. And vibration compensation is carried out in the measurement process, so that the accuracy and the stability of a measurement result are ensured.

Description

Technical Field

[0001] The invention belongs to the field of three-dimensional shape measurement and calibration, and in particular relates to a steel plate shape measurement and calibration method. Background Art

[0002] With the development of modern science and technology, rail transit, heavy equipment, military industry, aerospace and other fields have put forward higher requirements for the shape accuracy of plates. Straighteners and flattening machines are the main equipment on the plate finishing production line, and plate shape data is the basis for the intelligent realization of plate correction process. However, the specifications of plates are diverse, which poses a challenge to the plate shape measurement system. With the rapid development of visual technology in recent years, the structured light measurement technology characterized by simple structure, high efficiency and non-contact has also been rapidly improved. Building a multi-camera structured light measurement system is an important way to obtain the plate shape data of multi-specification plates, and it is also one of the key technologies for the intelligentization of metal rolling finishing equipment. The structured light measurement system is generally composed of a camera, a laser, a fixture, an industrial computer and related algorithms. Its measurement principle is mainly to collect the deformation data of the laser line hitting the surface of the object through the camera, and calculate the surface three-dimensional point cloud data of the measured object. In the structured light measurement system, the calibration method and accuracy of the measurement system directly affect the practicality and measurement accuracy of the system.

[0003] The calibration process of the current multi-camera, large-field-of-view structured light measurement system is relatively complicated, and most calibration methods have multiple fitting processes, which will cause the calibration accuracy to accumulate errors, limiting the accuracy of structured light measurement under a large field of view. At the same time, due to the vibration factors in the steel plate production process, the accuracy of structured light plate shape measurement is seriously affected. To this end, this patent proposes a new steel plate shape calibration and measurement method, which solves the problem of three-dimensional measurement of the plate shape of steel plates of different specifications, and solves the problem of the influence of plate vibration on plate shape measurement. The calibration process is simple and easy to implement on site. The structured light measurement system is mainly composed of multiple industrial cameras, two line lasers, a fixture, an industrial computer and related software. The combination of multiple cameras can meet the shape measurement needs of plates of different specifications. Summary of the invention

[0004] In order to solve the problem of low plate shape measurement accuracy, the present invention provides a steel plate shape measurement calibration method. The present invention combines multiple cameras and dual-line lasers, as well as vibration compensation technology, to achieve accurate measurement of the surface shape of the steel plate, providing reliable data support for quality control and process improvement in the plate production process.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a steel plate shape measurement and calibration method, including a calibration device and a measuring device, the calibration device includes a lifting device, a fixing device, a rotating device and a calibration plate, the calibration plate is installed on the upgrading device through the rotating device and can move with the lifting device, and the upgrading device is installed on both sides of the roller through the fixing device;

[0006] The measuring device includes multiple cameras and two line lasers. All cameras are fixed on the same horizontal plane at a certain distance, and the two lasers are placed in parallel and at a certain distance from each other.

[0007] The calibration and measurement process is carried out according to the following steps:

[0008] S1. Place the calibration device on the roller table of the steel plate shape measurement system and fix it, and determine that the position where the upper surface of the calibration plate is flush with the surface of the steel plate to be measured is the only reference surface. At this time, the working distance of the camera lens is H0;

[0009] S2, lift the calibration plate to a height of ΔH by the lifting platform of the calibration device, at which time the working distance of the camera lens is H2, and then calculate the single pixel accuracy P0 of the camera, the difference D0 between the laser center line coordinates and the image height center coordinates, the single pixel accuracy P2 of the camera, and the difference D2 between the laser center line coordinates and the image height center coordinates when the working distance of the camera lens is H0 and H2 respectively;

[0010] S3, using dual-line laser to perform three-dimensional reconstruction of the steel plate surface, the plate shape data represents the steel plate surface in the world coordinate system (X w , Y w , Z w )coordinate;

[0011] S31, Z of steel plate surface w Coordinate acquisition

[0012] According to the calibration process and the inherent properties of the camera and lens, P0, P2, D0, D2, v0, v2, image height V, image width U and lens field of view β are obtained. During the measurement process, the working distance of the camera lens is defined as H x , by establishing ΔH x (ΔHx=H x -H0) and the relationship between these parameters to solve for ΔH x , that is, Z w ;

[0013] During the measurement process, the steel plate is at different heights to be measured, and the position of the laser center line changes accordingly. This situation is divided into five cases. In each case, the triangular similarity principle is used to calculate ΔH x Establishing relationships with calibration parameters;

[0014] S32, X on the surface of steel plate w coordinate

[0015] The three-dimensional data calculated by each camera are stitched together. When the camera is installed, the installation position of each camera is determined so that each camera has a certain common field of view. Several calibration objects are placed on the calibration plate, and each calibration object is located within the common field of view of two adjacent cameras. Matching points are found through the pictures taken by the two cameras to obtain the space transformation matrix. Then, the coordinate systems of multiple cameras are mapped to the reference coordinate system by using the uniqueness of the world coordinate system of the calibration plate with the same posture in the common field of view of adjacent cameras. The coordinate systems of multiple cameras are unified into the same posture according to the uniqueness of the actual laser plane, and the positional relationship of the point cloud data of adjacent cameras is obtained after the unified coordinates.

[0016] S33, Y on the steel plate surface w coordinate

[0017] Y w According to the moving speed and time of the steel plate, take the running time t = 0 as the starting point, according to the moving speed V p , Y w The values ​​are:

[0018] Y w =V p *t.

[0019] Furthermore, the specific calculation process of step S2 is as follows:

[0020] S21. Get P0

[0021] There is a calibration object on the calibration board below the field of view of each two adjacent cameras. The side length of the first square in the calibration object is a in the real world. Then take a picture, and then use the program to detect the corner points of the black squares. By calculating the difference Δy of the v coordinates of the corner points in the image pixel coordinate system (that is, the number of coordinates occupied by the side length of a square in the image pixel coordinate system), the single pixel accuracy P0, that is, a / Δy, can be obtained;

[0022] S22, get D0

[0023] First, the grayscale centroid method in the laser centerline extraction algorithm is used to obtain the laser centerline coordinate v0 under H0. Then, given the image height V, its center coordinate is V / 2. Therefore, D0 = v0-V / 2 can be obtained.

[0024] S23. Using the same method, P2 and D2 when the working distance is H2 can be obtained.

[0025] Furthermore, in the five different cases in step S31, ΔH xThe relationship with the calibration parameters is expressed by the following formula:

[0026] (a) When v2 < V / 2, v x < V / 2, v0 < V / 2

[0027]

[0028] (b) When v2 < V / 2, v x < V / 2, v0 > V / 2

[0029]

[0030] (c) When v2 < V / 2, v x = V / 2, v0 > V / 2

[0031]

[0032] (d) When v2 < V / 2, v x > V / 2, v0 > V / 2

[0033]

[0034] (e) When v2 > V / 2, v x > V / 2, v0 > V / 2

[0035]

[0036] In the above formulas, the parameter ΔH = H2 - H0, and its value is directly obtained on the digital display of the calibration device. D x During the measurement process, by extracting the coordinates of the laser center line v x P0, D0, P2, and D2 are obtained through the calibration steps S1 and S2. ΔH x = H x - H0. In the above formula, H x is unknown, and P x is unknown;

[0037] According to the relationship between the parameters of the camera lens, formula (6) can be obtained

[0038]

[0039] According to formula (6), a relationship between the two unknowns H x , P x is established and联立 with formulas (1) to (5) respectively, and then ΔH in five cases can be obtained x , that is, Z w ;

[0040] During the measurement process, according to the position of the light strip pixels on the obtained image vx , determine which situation it is in, and then, except for situation (c) where the result is obtained directly, it is necessary to use equation (6) to calculate Z' w =ΔH x .

[0041] Furthermore, the measured Z w Coordinates for vibration compensation,

[0042] The measurement system has two laser lines R1 and R2, then at time t the two laser lines are calculated to get Z w The value can be expressed by equations (7) and (8),

[0043] Z' wR1 (t) = ΔH xR1 #(7)

[0044] Z' wR2 (t) = ΔH xR2 #(8)

[0045] When the measurement time interval is set to the time interval when the steel plate moves from R1 to R2, the R1 laser line will measure the same position as the R2 laser line at the previous moment every Δt. In this way, the steel plate position measured by R2 at time t1 is the same as the steel plate position measured by R1 at time t2, and the results should also be the same. However, due to vibration, the deviation between the two is S1.

[0046] S1=Z' wR2 (t1)-Z' wR2 (t2)#(9)

[0047] Add the vibration amount S1 to all the measurement results at time t2 and after to eliminate the vibration, and so on. i The measurement result of the position at time R1 will produce S compared with the measurement result of the position at the previous time R2. i-1 The vibration value can be positive or negative. If the first measurement position is set at R2 at time t1, then at t i When measuring the steel plate at any time, the height Z after eliminating vibration w The coordinates should be:

[0048]

[0049] Furthermore, the X on the surface of the steel plate w The specific calculation process of the coordinates is:

[0050] P iH is the center coordinate of the landmark on the right side of the i-th camera, P iL is the center coordinate of the landmark on the left side of the i-th camera, let ΔL i is the actual length of the i-th calibration object, Δvi is the number of pixels occupied by the size of the calibration object on the left side of the i-th camera in the image, Δv' i is the number of pixels occupied by the size of the calibration object on the right side of the i-th camera in the image, P i is the actual size represented by the unit pixel of the ith camera, according to the following formula (6), and substitute the working distance H of the ith camera based on the reference plane i , we can get:

[0051]

[0052] ΔH i It indicates the height of the area when stitching. The coordinates increase to the right with the first camera as the reference. Formula (12) indicates the coordinate calculation of the board shape 3D data in the board width direction when there are i cameras:

[0053] X w =P 1H *P1+(P 2H -P 2L )*P2+…+(ν i -P iL )*P i v i ≥P iL ,i≥1#(12).

[0054] The advantages and positive effects of the present invention are as follows: the present invention mainly utilizes the designed calibration device to quantify the positional relationship of the laser line in the camera, and considers the difference in the actual size represented by the pixel at different distances between the target object and the camera in the calculation of the three-dimensional data, thereby compensating for the influence of the undulating shape of the plate on the measurement error. At the same time, according to the vibration compensation calculation method, the influence of vibration on the three-dimensional measurement is weakened. Finally, according to the relative positional relationship of each camera, the point cloud data of multiple cameras are spliced ​​and calculated, and the multi-camera measurement data are unified under the reference coordinates, thereby realizing the measurement of the plate shape data of the standard plate. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is the front view of the calibration device of the present invention.

[0056] Figure 2 Top view of the calibration device.

[0057] Figure 3 This is the left view of the calibration device.

[0058] Figure 4 Schematic diagram of the measurement system based on multiple cameras.

[0059] Figure 5 Schematic diagram of the calibration principle.

[0060] Figure 6 Calculate schematics for flatness data.

[0061] Figure 7 Schematic diagram of the relationship between field of view angle and working distance.

[0062] Figure 8 This is the schematic diagram of vibration compensation calculation.

[0063] Fig. 9 This is a schematic diagram of three-dimensional data stitching. DETAILED DESCRIPTION

[0064] In order to make the objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below.

[0065] A steel plate shape measurement and calibration method mainly includes two parts: calibration and measurement. Each step requires a corresponding hardware device. First, a brief introduction is given to the hardware device involved in the present invention.

[0066] like Figure 1-Figure 3 As shown in the figure, the calibration device mainly consists of four parts: lifting device, fixing device, rotating device and calibration plate. First, the lifting device adopts a high-precision, heavy-load digital display lifting platform. To ensure accuracy, the lifting height accuracy of the lifting platform is not less than 0.05 mm, and the reading accuracy of the digital display device installed on the side is 0.01 mm to obtain the precise value of the lifting height of the calibration plate.

[0067] Secondly, the fixture is built of aluminum profiles in a rectangular frame with fixed plates on both sides for fixing the lifting platform. The rotating device is designed to prevent the calibration plate from deforming due to the inconsistent heights on both sides during the lifting process. In addition, in order to increase the bending strength, two aluminum profiles are welded on the lower surface of the calibration plate to resist deformation caused by gravity. A checkerboard calibration plate is placed on its upper surface for calculating the calibration parameters.

[0068] like Figure 4 As shown in the figure, the steel plate shape measurement system includes multiple cameras and two line lasers. All cameras are fixed on the same horizontal plane at a certain distance, while the two lasers are placed in parallel and at a certain distance from each other. The number of cameras can be determined according to the specifications of the plate.

[0069] The above briefly describes the specific hardware structure of the present invention. The following briefly introduces the calibration principle of the present invention to facilitate those skilled in the art to understand the calibration measurement method of the present invention.

[0070] like Figure 5As shown, the calibration method proposed by the present invention is to deduce the three-dimensional height of the steel plate surface based on the change of the position of the laser line in the image at the image height. Therefore, the main purpose of calibration is to determine the conversion relationship between pixels and height.

[0071] In order to obtain height information from the laser line position in the image, this paper proposes a pixel offset direct calibration method. First, the camera imaging model is simplified to a projection model. Figure 5 According to the relationship shown in , the triangle projection formed by the camera optical center on any column v parallel to the pixel coordinate u direction is OAB. On the OAB plane, when the height changes by ΔH, the target object will produce an offset of Δδ pixels on the image plane under the irradiation of the laser. Based on the above relationship, we proposed a calibration method and designed a calibration device, such as Figures 1 to 3 As shown, the device can not only move the calibration plate up and down freely but also display the movement amount of the calibration plate in real time.

[0072] When the calibration plate moves up and down, the laser line coordinates produce a pixel offset Δδ, and the working distance H of the camera lens will also change, which will affect the field of view and cause changes in single pixel accuracy. However, in the measurement process, in order to improve accuracy, this effect needs to be reduced. Therefore, during the calibration process, the single pixel accuracy P at different working distances must be calculated. i , and substitute it into the derivation process of the three-dimensional steel plate surface height.

[0073] It can be seen that the calibration process is the process of determining the following parameters

[0074] H i : The working distance of the camera lens, that is, the distance between the lens and the object being measured.

[0075] P i : The camera lens working distance is H i When the camera captures an image, the single pixel accuracy of the camera (that is, the actual size represented by a pixel in the image captured by the camera) is used.

[0076] D i : The camera lens working distance is H i The difference between the laser center line coordinates and the image height center coordinates when .

[0077] v i : The camera lens working distance is H i The V direction coordinate in the pixel coordinate system of the laser centerline image when .

[0078] Based on the above hardware structure and calibration principle, the specific calibration measurement process of the present invention is introduced in detail.

[0079] S1. Place the calibration device on the roller in the steel plate shape measurement system and fix it. Then determine that the position where the upper surface of the calibration plate is flush with the surface of the steel plate to be measured is the only reference surface, and the working distance of the camera lens is H0.

[0080] S2. Lift the calibration plate to a height of ΔH through the lifting platform of the calibration device. At this time, the working distance of the camera lens is H2. Calculate P0, D0, P2, D2 at the two working distances respectively. The specific steps are as follows:

[0081] S21. Get P0

[0082] Below the field of view of each two adjacent cameras, there is a calibration object on the calibration board. The side length of the first square in the calibration object in the real world is a (as shown above). Figure 2 Then, the program is used to detect the corners of the black squares, and by calculating the difference Δy of the v coordinates of the corners in the image pixel coordinate system (that is, the number of coordinates occupied by the side length of a square in the image pixel coordinate system), the single pixel accuracy P0, that is, a / Δy, can be obtained.

[0083] S22, get D0

[0084] First, the grayscale centroid method in the laser centerline extraction algorithm (which has the advantages of fast speed, high accuracy and can achieve sub-pixel accuracy level centerline extraction) is used to obtain the laser centerline coordinate v0 under H0. Then, given the image height V, its center coordinate is V / 2. Therefore, D0 = v0-V / 2 can be obtained.

[0085] Using the same method, P2 and D2 when the working distance is H2 can be obtained.

[0086] S3. Steel plate shape measurement is to perform three-dimensional reconstruction of the steel plate surface. Its shape data represents the steel plate surface in the world coordinate system (X w , Y w , Z w ) coordinates, the present invention uses a dual-line laser to perform three-dimensional reconstruction of the steel plate surface (as shown above Figure 4 shown).

[0087] S31, Z of steel plate surface w coordinate

[0088] According to the calibration process and the inherent properties of the camera and lens, P0, P2, D0, D2, v0, v2, image height V, image width U and lens field of view β can be obtained. During the measurement process, the working distance of the camera lens is defined as H x . (i.e. the distance between the steel plate to be tested and the camera lens), by establishing ΔH x (ΔH x =H xSolve for ΔH based on the relationship between ΔH x , that is, Z w .

[0089] During the measurement process, the steel plate is at different heights to be measured, and the position of the laser center line also changes accordingly. This situation is divided into five cases. In each case, the principle of similar triangles is used to establish the relationship between ΔH x and the calibration parameters. As Figure 6 shown.

[0090] According to Figure 8 the similar triangle relationships in (a) to (e) below, the following formulas can be obtained:

[0091] (a) When v2 < V / 2, v x < V / 2, v0 < V / 2

[0092]

[0093] (b) When v2 < V / 2, v x < V / 2, v0 > V / 2

[0094]

[0095] (c) When v2 < V / 2, v x = V / 2, v0 > V / 2

[0096]

[0097] (d) When v2 < V / 2, v x > V / 2, v0 > V / 2

[0098]

[0099] (e) When v2 > V / 2, v x > V / 2, v0 > V / 2

[0100]

[0101] In the above formulas, the parameter ΔH = H2 - H0, and its value can be directly obtained on the digital display of the calibration device. D x can be obtained by extracting the laser center line coordinates v x during the measurement process. P0, D0, P2, and D2 have been obtained through calibration. ΔH x = H x - H0, where H x is unknown, and P x is unknown.

[0102] In addition, according to the relationship between the parameters of the camera lens, Figure 7 .

[0103] In the figure, W0 and W2 are the field of view at working distances H0 and H2. β is the vertical field of view of the lens.

[0104] According to the schematic diagram, formula (6) can be obtained.

[0105]

[0106] According to this formula, we can establish two unknown quantities H x , P x The relationship between and can be combined with equations (1) to (5) to obtain ΔH in five cases. x , that is, Z w .

[0107] During the measurement process, the pixel position v of the light bar on the obtained image is x , determine which situation it is in, and then, except for situation (c) where the result is obtained directly, it is necessary to use equation (6) to calculate Z' w =ΔH x .

[0108] Since the plate shape detection system is an online real-time detection, the steel plate will vibrate during transportation, which will lead to large errors in the detection results. Therefore, it is necessary to compensate for the vibration of the measurement results. The vibration is mainly vertical vibration, so only the Z w The following is a vibration compensation method and principle.

[0109] When the steel plate is measured at a constant speed and equal distance, R1 and R2 are used to represent two laser lines respectively. Then, at time t, the two laser lines are calculated to obtain Z w The values ​​can be expressed by equations (7) and (8).

[0110] Z' wR1 (t) = ΔH xR1 #(7)

[0111] Z' wR2 (t) = ΔH xR2 #(8)

[0112] Every time Δt passes, the steel plate at the R2 laser line will move to the R1 position, which is equivalent to the laser line moving backward on the steel plate for measurement. When the measurement time interval is set to the time interval for the steel plate to move from R1 to R2, every Δt will make the R1 laser line measure the same position as the R2 laser line at the previous moment. In this way, the steel plate position measured by R2 at time t1 is the same as the steel plate position measured by R1 at time t2, and the results should also be the same, but due to vibration, the two will have a deviation S1, such as Figure 8 Shown by red dashed line.

[0113] S1=Z' wR2 (t1)-Z' wR2 (t2)#(9)

[0114] Therefore, all the measurement results at and after t2 are added with the vibration amount S1 to eliminate the vibration. i The measurement result of the position at time R1 will produce S compared with the measurement result of the position at the previous time R2. i-1 The vibration value can be positive or negative. If the first measurement position is set to R2 at time t1, then at t i When measuring the steel plate at any time, the height Z after eliminating vibration w The coordinates should be:

[0115]

[0116] S32, X on the surface of steel plate w coordinate

[0117] The measurement system consists of multiple cameras. In order to complete the entire plate shape data measurement, the 3D data calculated by each camera needs to be spliced. When installing the camera, the installation position of each camera is determined so that each camera has a certain common field of view, such as Fig. 9 As shown, several calibration objects are placed on the calibration plate, and each calibration object is located within the common field of view of two adjacent cameras. Matching points are found through the images taken by the two cameras to obtain the spatial transformation matrix. Then, the coordinate systems of multiple cameras are mapped to the reference coordinate system by utilizing the uniqueness of the world coordinate system of the calibration plate with the same posture in the common field of view of adjacent cameras. According to the uniqueness of the actual laser plane, the coordinate systems of multiple cameras are unified into the same posture. The positional relationship of the point cloud data of adjacent cameras after the unified coordinates is used, and the point cloud data is directly spliced ​​according to the coordinate mapping result, which effectively avoids the large amount of calculation brought by using point cloud data for feature matching and realizes the rapid splicing of multi-camera point cloud data.

[0118] Fig. 9 In, P iH is the center coordinate of the landmark on the right side of the i-th camera, P iL is the center coordinate of the marker on the left side of the i-th camera. Let ΔL i is the actual length of the i-th calibration object, Δv i is the number of pixels occupied by the size of the calibration object on the left side of the i-th camera in the image, Δv' i is the number of pixels occupied by the size of the calibration object on the right side of the i-th camera in the image. iis the actual size represented by the unit pixel of the i-th camera. According to formula (6), and substituting the working distance H of the i-th camera based on the reference plane i , we can get:

[0119]

[0120] ΔH i Indicates the height of the area when stitching. Taking the first camera as the reference, the coordinates increase to the right. Formula (12) indicates the coordinate calculation of the board shape 3D data in the board width direction when there are i cameras in total.

[0121] X w =P 1H *P1+(P 2H -P 2L )*P2+…+(v i -P iL )*P i v i ≥P iL ,i≥1#(12)

[0122] S33, Y on the steel plate surface w coordinate

[0123] Y w According to the moving speed and time of the steel plate, take the running time t = 0 as the starting point, according to the moving speed V p , Y w The values ​​are:

[0124] Y w =V p *t#(13)

[0125] The present invention realizes three-dimensional reconstruction and shape measurement of the steel plate surface. In order to realize high-precision detection of the plate shape, the present invention obtains X, Y, and Z coordinate data of the steel plate surface through a dual-line laser measurement system.

[0126] The embodiments of the present invention are described in detail above, but the present invention is not limited to the above embodiments. Various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A steel plate shape measurement and calibration method, characterized in that: It includes a calibration device and a measuring device. The calibration device includes a lifting device, a fixing device, a rotating device and a calibration plate. The calibration plate is installed on the upgrading device through the rotating device and can move with the lifting device. The upgrading device is placed on both sides of the roller table through the fixing device. The measuring device includes multiple cameras and two line lasers. All cameras are fixed on the same horizontal plane at a certain distance, and the two lasers are placed in parallel and at a certain distance from each other. The calibration measurement process is carried out according to the following steps: S1. Place the calibration device on the roller table of the steel plate shape measurement system and fix it, and determine that the position where the upper surface of the calibration plate is flush with the surface of the steel plate to be measured is the only reference surface. At this time, the working distance of the camera lens is H0; S2, lift the calibration plate to a height of ΔH by the lifting platform of the calibration device, at which time the working distance of the camera lens is H2, and then calculate the single pixel accuracy P0 of the camera, the difference D0 between the laser center line coordinates and the image height center coordinates, the single pixel accuracy P2 of the camera, and the difference D2 between the laser center line coordinates and the image height center coordinates when the working distance of the camera lens is H0 and H2 respectively; S3, using dual-line laser to perform three-dimensional reconstruction of the steel plate surface, the plate shape data represents the steel plate surface in the world coordinate system (X w , Y w , Z w )coordinate; S31, Z of steel plate surface w Coordinate acquisition According to the calibration process and the inherent properties of the camera and lens, P0, P2, D0, D2, v0, v2, image height V, image width U and lens field of view β are obtained. During the measurement process, the working distance of the camera lens is defined as H x , by establishing ΔH x (ΔHx=H x -H0) and the relationship between these parameters to solve for ΔH x , that is, Z w ; During the measurement process, the steel plate is at different heights to be measured, and the position of the laser center line changes accordingly. This situation is divided into five cases. In each case, the triangular similarity principle is used to calculate ΔH x Establishing relationships with calibration parameters; S32, X on the surface of steel plate w coordinate The three-dimensional data calculated by each camera are stitched together. When the camera is installed, the installation position of each camera is determined so that each camera has a certain common field of view. Several calibration objects are placed on the calibration plate, and each calibration object is located within the common field of view of two adjacent cameras. Matching points are found through the pictures taken by the two cameras to obtain the space transformation matrix. Then, the coordinate systems of multiple cameras are mapped to the reference coordinate system by using the uniqueness of the world coordinate system of the calibration plate with the same posture in the common field of view of adjacent cameras. The coordinate systems of multiple cameras are unified into the same posture according to the uniqueness of the actual laser plane, and the positional relationship of the point cloud data of adjacent cameras is obtained after the unified coordinates. S33, Y on the steel plate surface w coordinate Y w According to the moving speed and time of the steel plate, starting from the running time t = 0, according to the moving speed V p , Y w The values ​​are: Y w =V p *t。 2. A steel plate shape measurement and calibration method according to claim 1, characterized in that: The specific calculation process of step S2 is as follows: S21. Get P0 There is a calibration object on the calibration board below the field of view of each two adjacent cameras. The side length of the first square in the calibration object is a in the real world. Then take a picture, and then use the program to detect the corner points of the black squares. By calculating the difference Δy of the v coordinates of the corner points in the image pixel coordinate system (that is, the number of coordinates occupied by the side length of a square in the image pixel coordinate system), the single pixel accuracy P0, that is, a / Δy, can be obtained; S22, get D0 First, the grayscale centroid method in the laser centerline extraction algorithm is used to obtain the laser centerline coordinate v0 under H0. Then, given the image height V, its center coordinate is V / 2. Therefore, D0 = v0-V / 2 can be obtained. S23. Using the same method, P2 and D2 when the working distance is H2 can be obtained.

3. A steel plate shape measurement and calibration method according to claim 1, characterized in that: In the five different cases in step S31, ΔH x The relationship with the calibration parameters is expressed by the following formula: (a) When v2 < V / 2, v x < V / 2, v0 < V / 2 (b) When v2 <V / 2,v x <V / 2,v0> V / 2 (c) When v2 <V / 2,v x =V / 2,v0>V / 2 (d) When v2 <V / 2,v x >V / 2,v0>V / 2 (e) When v2 > V / 2, v x > V / 2, when v0 > V / 2 In the above formulas, the parameter ΔH = H2-H0, and its value is directly obtained on the digital display of the calibration device, D x During the measurement process, the laser center line coordinate v is extracted x Obtain, P0, D0, P2, D2 have been obtained through calibration steps S1, S2, ΔH x =H x -H 0, , in the above formula, H x Unknown, P x unknown; According to the relationship between the parameters of the camera lens, we can get formula (6): According to formula (6), two unknown quantities H are established x , P x The relationship between and can be combined with equations (1) to (5) to obtain ΔH in five cases. x , that is, Z w ; During the measurement process, the pixel position v of the light bar on the obtained image is x , determine which situation it is in, and then, except for situation (c) where the result is obtained directly, it is necessary to use equation (6) to calculate Z' w =ΔH x .

4. A steel plate shape measurement and calibration method according to claim 3, characterized in that: The measured Z w Coordinates for vibration compensation, The measurement system has two laser lines R1 and R2, then at time t the two laser lines are calculated to get Z w The value can be expressed by equations (7) and (8), Z′ wR1 (t)=ΔH xR1 #(7) Z′ wR2 (t)=ΔH xR2 #(8) When the measurement time interval is set to the time interval when the steel plate moves from R1 to R2, the R1 laser line will measure the same position as the R2 laser line at the previous moment every Δt. In this way, the steel plate position measured by R2 at time t1 is the same as the steel plate position measured by R1 at time t2, and the results should also be the same. However, due to vibration, the deviation between the two is S1. S1=Z′ wR2 (t1)-Z′ wR2 (t2)#(9) Add the vibration amount S1 to all the measurement results at time t2 and after to eliminate the vibration, and so on. i The measurement result of the position at time R1 will produce S compared with the measurement result of the position at the previous time R2. i-1 The vibration value can be positive or negative. If the first measurement position is set at R2 at time t1, then at t i When measuring the steel plate at any time, the height Z after eliminating vibration w The coordinates should be:

5. A steel plate shape measurement and calibration method according to claim 4, characterized in that: The X w The specific calculation process of the coordinates is: P iH is the center coordinate of the landmark on the right side of the i-th camera, P iL is the center coordinate of the landmark on the left side of the i-th camera, let ΔL i is the actual length of the i-th calibration object, Δv i is the number of pixels occupied by the size of the calibration object on the left side of the i-th camera in the image, Δv' i is the number of pixels occupied by the size of the calibration object on the right side of the i-th camera in the image, P i is the actual size represented by the unit pixel of the ith camera, according to the following formula (6), and substitute the working distance H of the ith camera based on the reference plane i , we can get: ΔH i It indicates the height of the area when stitching. The coordinates increase to the right with the first camera as the reference. Formula (12) indicates the coordinate calculation of the board shape 3D data in the board width direction when there are i cameras: X w =P 1H *P1+(P 2H -P 2L )*P2+…+(in i -P iL )*P i in i ≥P iL ,i≥1#(12).

Images