A complete calibration method for line structured light vision sensors based on single cylindrical targets

Abstract

The present invention proposes a complete calibration method for a line structured light system based on a single cylindrical target. The calibration process is as follows: the cylindrical target is placed at least once and illuminated with a light strip of a laser projector. The visual sensor captures the target image and extracts the laser stripes and contour image of the cylindrical target. According to the contour features of the cylindrical target and the definition of the collinear vector, the extracted contour image is used to solve the internal parameters of the camera. Since the intersection of the line structured light plane and the cylindrical target in the projection direction of the cylinder diameter is a standard circle, the light plane equation is established in the camera coordinate system. In a complex industrial environment, it is inappropriate to use the traditional Zhang calibration method to shoot multiple chessboard images at different positions, and due to the contour characteristics of the cylindrical target, camera calibration and line structure calibration can be performed simultaneously. The method proposed in the present invention does not need to rely on the constraints of multiple images, is simple to operate, and meets the requirements of industrial three-dimensional measurement technology.

Description

Technical Field

[0001] The present invention belongs to the field of camera three-dimensional imaging and detection technology, and specifically relates to a visual inspection method that simultaneously performs camera calibration and line structured light plane calibration through a 3D cylindrical target to achieve real-time detection in a complex industrial environment. Background Art

[0002] Currently, there are numerous calibration methods for vision measurement systems, but most are conducted in laboratories. With advances in science and technology, industrial production is gradually transitioning from manual assembly lines to fully automated production, placing higher demands on measurement speed and accuracy in the industrial inspection field. The increasing demand for non-contact measurement in practical measurement applications has driven the shift of vision measurement technology from laboratory environments to complex industrial production environments. Many challenges remain in transitioning from laboratory to practical applications, such as increasing the measurement range of measurement systems and their ability to inspect more complex surfaces. Furthermore, it is important to select a convenient and appropriate calibration method based on the specific measurement scenario. Different calibration methods have different impacts on 3D measurement systems, and an inappropriate calibration method can easily lead to low calibration accuracy, thereby reducing the reliability of 3D measurement. Furthermore, linear, intelligent, and flexible measurement systems have also attracted considerable industry attention. Currently, there are numerous calibration methods for vision measurement systems, but most require complex algorithms with numerous and interrelated parameters. The reliability of the measurement process depends on a rigorous calibration environment, and these complex algorithms are time-consuming to calibrate. Consequently, most traditional vision measurement methods cannot meet the frequent, fast, and convenient calibration requirements of industrial sites. Over the years, people have done a lot of related research. Through the careful design of three-dimensional sensor systems, optimization of various system parameters, and good system calibration, the measurement accuracy of the measurement system can be improved by an order of magnitude. However, in general, the construction of the measurement system is still constrained by specific conditions in actual applications, and the complex calibration process makes the measurement system less operational in industrial production.

[0003] During camera calibration, traditional camera intrinsic parameters are primarily calibrated using two-dimensional (2D) or three-dimensional (3D) coplanar point calibration targets. While 2D coplanar point calibration targets are simple to fabricate, a single image is insufficient to obtain all camera parameters. Multiple images at different angles and positions are required, requiring nonlinear solutions based on the positional relationships between these images. While 3D calibration targets are more difficult to fabricate than 2D targets, under certain conditions, only a single image is needed to determine all camera intrinsic and extrinsic parameters. This method is suitable for industrial measurement. During online structured light calibration, calibration targets are categorized by target type: one-dimensional (1D), two-dimensional (2D), three-dimensional (3D), and virtual targets. 1D and 2D targets typically require multiple views at different angles and poses to complete structured light calibration. One-dimensional targets are primarily used for on-site calibration with a large field of view. 3D targets utilize the geometric information provided by the target itself to perform light plane calibration within a single view. Virtual targets utilize the relative nature of motion to control the translation of the structured light vision sensor for light plane calibration. Although calibration can be performed without the use of additional targets, it relies on the motion information of the sensor.

[0004] It can be seen that there is still a lot of room for improvement and application progress in the calibration method of line structured light vision sensor measurement system based on three-dimensional targets. Summary of the Invention

[0005] This paper proposes a new calibration method that utilizes the outer contour characteristics of a single cylindrical 3D target to simultaneously perform camera calibration and line structured light plane calibration. This method improves the real-time performance and simplicity of existing calibration methods in complex industrial environments, achieving the requirement for simultaneous calibration of a single image. This method is achieved through the following technical solutions:

[0006] 1. A complete calibration method for a line structured light vision sensor based on a single cylindrical target, comprising:

[0007] A laser transmitter, a camera, a tripod and a three-dimensional target. The three-dimensional target is in front of the laser transmitter and within the field of view of the camera. The camera is fixed on the tripod.

[0008] The three-dimensional target is a cylindrical target.

[0009] The line structured light emitted by the laser emitter is located on the cross section of the single cylindrical target.

[0010] The intersection line between the line structured light emitted by the laser emitter and the cross section of the single cylindrical target is a standard circle in the axial projection and an ellipse in the radial projection.

[0011] The conversion function corresponding to the camera calibration process can be expressed as:

[0012] (1)

[0013] in, ρ is a non-zero scale factor; A is the camera internal parameter matrix, which is only related to the camera internal parameters. u 0 and v 0 represents the coordinates of the camera's optical center. a x and a y The images are u and v Normalized focal length on axis; R is the rotation matrix of the image plane, t is the three-dimensional translation vector; M is the camera's external parameter matrix, which is determined by the camera's orientation to the world coordinate system.

[0014] Establish the world coordinate system in the camera coordinate system. O c -X c Y c Z c Represents the camera coordinate system.

[0015] (2)

[0016] 2. A complete calibration method for a line structured light vision sensor based on a single cylindrical target, characterized by comprising the following steps.

[0017] Step 1: The cylindrical target is placed in a suitable location. The camera captures the edge information of the cylindrical target and the light stripes at the intersection of the line structured light vision sensor and the cylindrical target. It is worth noting that due to the unique geometric characteristics of the cylindrical target, any position of the cylindrical target is allowed at this point. The only requirement is to distinguish the cylindrical target from the external environment color.

[0018] Step 2: By ensuring that the modulus of the outer product of collinear vectors is zero and the distance constraint of cylindrical symmetry is used, the camera distortion parameters are solved. The camera focal length is then solved using the depth information of the cylindrical target edge. After obtaining the camera's internal parameters, the image is subjected to distortion correction.

[0019] Step 3: Extract the center stripes of the light after distortion correction, solve the spatial cone equation, and use the projection properties combined with the cylindrical target information to obtain the light plane equation.

[0020] according to Fitzgibbon The proposed single-division lens distortion model denotes the ideal imaging point as ( u p , v p), the point distorted by the lens is recorded as ( u d , v d ), there is the following relationship between the two:

[0021] (3)

[0022] in( u 0, v 0) is the center of distortion, ,coefficient k 1 Represents the first-order radial distortion of the lens.

[0023] For nonlinear camera lens distortion correction, the camera lens distortion can be solved by using the available information of a given scene before calibration.

[0024] According to geometry, if two vectors are collinear, their outer product is a zero vector. Therefore, for any two points on a straight line, the modulus of their outer product is zero due to collinearity. In this sense, the modulus of the outer product of collinear vectors can be used as a distortion measure to correct lens distortion. For the selected calibration target, the feature points of two straight lines can be extracted on the calibration image, and each straight line has N n ( n =1,2) feature points. Examine two adjacent feature points P i and P i+1 The vector P i P i+1 , P i is the ideal imaging point. According to the vector sequence constructed by this method, take two adjacent vectors for cross product operation and calculate their modulus, and we get:

[0025] (4)

[0026] According to the symmetry of the cylinder, the distances between symmetrical points are equal. For the selected calibration target, the feature points of two curves can be extracted on the calibration image. Each curve is divided into two symmetrical parts, each part has N m ( m =1,2,3,4) feature points. According to the corresponding feature points D m,j ( j =1,2,…, N m ) (in Dm,j is the spatial coordinate in the camera coordinate system), we can get:

[0027] (5)

[0028] According to the perspective projection model, we can simplify it and get:

[0029] (6)

[0030] Where dx and dy represent the physical size of each pixel in the x-axis and y-axis directions respectively;

[0031] In order to overcome the influence of noise, the edge is fitted with a curve, and the curve circle usually used is a quadratic curve. Substitute the distortion model, that is, equation (3), into equations (4) and (6).

[0032] Camera first-order radial distortion coefficient k 1 and the center of lens distortion ( u 0, v The specific implementation steps of the parameter estimation algorithm of 0) are as follows:

[0033] (1) Determine the range of distortion coefficient and distortion center. is the maximum distance from the distortion center to the edge of the image, we can get k The range of 1 is . Assume the image size is w × h ( w and h are the length and width of the image respectively), and the center of distortion is generally located near the center of the image 0.1 w ×0.1 h In a rectangular area of size , .

[0034] (2) Zernike moments are used to extract sub-pixel features of the cylindrical target edge contour to obtain the edge sub-pixel points of the original distorted image. Since structured light fringes have a certain influence on image edge extraction, this paper combines the lens distortion model to perform quadratic curve fitting on the feature points of the two straight lines extracted from the selected calibration target on the calibration image.

[0035] (3) According to the above distortion parameters k 1, u 0, v 0, and select the corresponding step size Δ k 1, Δ u 0,Δ v 0, get the distortion parameter combination

[0036]

[0037] in , , , , , ; For each coefficient combination in the formula Computes the correction of the edge image.

[0038] (4) Use nonlinear optimization to search for the various coefficient combinations mentioned above to obtain the optimal solution, and the corresponding parameters are the optimal values.

[0039] After the complete distortion correction of the image, the focal length of the image can be obtained according to the geometric information of the cylindrical target. After the distortion correction of the feature points of the two straight lines extracted from the calibration target selected above, there is still N n ( n =1,2) feature points. Since the cylindrical target may not be placed completely horizontally, the present invention establishes a new coordinate system rotated by θ degrees around the z-axis of the camera coordinate system, and rotates the obtained target image into an image completely perpendicular to the horizontal plane. According to the properties of rigid body transformation, it can be obtained:

[0040] (7)

[0041] Using the perspective projection model we can get:

[0042] (8)

[0043] k is the depth ratio of the focal length to the central axis of the cylinder, that is k=f / z 1; d is the height of the cylindrical target; x 1 is the cylindrical target point P 3,4 In the camera coordinate system x Axis position; y 1 is the cylindrical target point P 3,4 In the camera coordinate system y The position of the axis.

[0044] Using nonlinear least squares algorithm to obtain x 1, y 1, k The curve fitted after distortion correction is an ellipse and xoz The projection on the plane is a perfect circle.

[0045] We can get:

[0046] (9)

[0047] According to the properties of the fitted ellipse, we can get:

[0048] (10)

[0049] in, s is any non-zero scale, t 1 The center of the cylindrical target x Axis information and t 1 =x 1 +Rx , t 2 is the center of the cylindrical target z Axis information, Rx is the radius of the cylindrical target. According to formula (10), the normalized focal length can be obtained a x 、 a y .

[0050] In the process of solving the light plane equation, the camera coordinate system ( x , y , z ), world coordinate system ( x w , y w , z w ), then the target plane equation in the camera coordinate system is:

[0051] (11)

[0052] in a , b , c , d Represent the four parameters of the light plane equation.

[0053] 3 According to Steger The algorithm extracts the central stripes of the image plane. The obtained central stripes are used to perform ellipse fitting to obtain C .

[0054] The cylindrical targets intersect to form an elliptical cone in three-dimensional space. According to the camera model, the ellipse fitted at the center of the light stripe is substituted into formula (2) to obtain:

[0055] (12)

[0056] in , assuming that the plane equation of the line structured light is: y=ax+cz+d , we can get the intersection line of the line structured light and the cylindrical target at xoz Projection equation on the surface.

[0057] (13)

[0058] Since the elliptical cone xoz The projection line on the plane is a circle and we get:

[0059] (14)

[0060] in, s is any non-zero scale. According to the properties, we can get a, c, d Therefore, the light plane equation is established in the camera coordinate system, and the coefficients of the light plane equation are:

[0061] (15)

[0062] This paper proposes a complete calibration method for a linear structured light vision sensor based on a single cylindrical target, based on the measurement requirements of a replicated industrial environment. The complete calibration method of the present invention has the following advantages:

[0063] (1) The cylindrical target used in the present invention is easy to process and has high processing accuracy. The cylindrical target also has certain advantages in camera internal parameter calibration.

[0064] (2) The present invention can complete the light plane calibration process with only one calibration image. The three-dimensional measurement of the camera has high requirements for the calibration site environment and strict control of the intermediate calibration parameters. The two methods proposed in the present invention have simple calibration processes and short calibration time, thus meeting the needs of industrial production that increasingly emphasizes real-time and online performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 Line structured light vision sensor target model.

[0066] Figure 2 Light stripes projected from line structured light onto a cylindrical target.

[0067] Figure 3 Light streaks after image processing.

[0068] Figure 4 Calibration principle between symmetrical points.

[0069] Figure 5 Fitted ellipse. DETAILED DESCRIPTION

[0070] The specific implementation of the present invention is described in detail with reference to the accompanying drawings.

[0071] like Figure 1 As shown in the figure, a calibration system for a line structured light vision sensor based on a single cylindrical 3D target includes the following four parts: a cylindrical 3D target 1 with a diameter of 100 mm and an accuracy of 0.02 mm. A single-line laser projector 2, a CCD camera 3 with BFS-U3-23S3C-C (resolution 1920 pixels x 1200 pixels) and a measuring distance of approximately 1200 mm. A tripod 4. The cylindrical target is in front of the laser emitter, and the three-dimensional cylindrical target is within the camera's field of view. The camera is fixed on the tripod. Figure 2 As shown in FIG, the intersection line of the line structured light emitted by the laser emitter and the cross section of the single cylindrical target is a standard circle in the axial projection and an ellipse in the radial projection. The calibration method of the system is applied, as shown in FIG. Figure 2 As shown, the world coordinate system and the camera coordinate system are established at the same position, and the camera calibration and laser plane calibration can be performed simultaneously.

[0072] The embodiment of the present invention is as follows: the line structured light vision sensor is composed of a camera and a laser projector and uses a CCD camera with BFS-U3-23S3C-C (resolution 1920 pixels x 1200 pixels) and a measuring distance of about 1200 mm to shoot the laser stripes projected by the laser projector on the cylindrical target, such as Figure 1 shown.

[0073] The calibration method proposed in this embodiment is used to calibrate the structured vision sensor. A cylindrical target is placed in front of the linear structure optical vision sensor. The diameter of the cylindrical target is 100 mm and the accuracy is 0.02 mm. The image used for this calibration is as follows Figure 3 shown.

[0074] According to the symmetry of the cylinder, the distances between symmetrical points are equal. For the selected calibration target, the feature points of two curves can be extracted on the calibration image. Each curve is divided into two symmetrical parts, each part has N m ( m =1,2,3,4) feature points. The calibration principle is as follows Figure 4 As shown, (taking 4 points as an example in the figure), according to the corresponding feature points D m,j ( j =1,2,…, N m ) (in D m,j is the spatial coordinate in the camera coordinate system), we can get:

[0075] (16)

[0076] Camera first-order radial distortion coefficient k 1 and the center of lens distortion ( u 0, v The specific implementation steps of the parameter estimation algorithm of 0) are as follows:

[0077] (1) Determine the range of distortion coefficient and distortion center. is the maximum distance from the distortion center to the edge of the image, we can get k The range of 1 is . Assume the image size is w × h ( w and h are the length and width of the image respectively), and the center of distortion is generally located near the center of the image 0.1 w ×0.1 h In a rectangular area of size , .

[0078] (2) Zernike moments are used to extract sub-pixel features of the cylindrical target edge contour to obtain the edge sub-pixel points of the original distorted image. Since structured light fringes have a certain influence on image edge extraction, this paper combines the lens distortion model to perform quadratic curve fitting on the feature points of the two straight lines extracted from the selected calibration target on the calibration image.

[0079] (3) According to the above distortion parameters k 1, u 0, v 0, and select the corresponding step size Δ k 1, Δ u 0,Δ v 0, get the distortion parameter combination

[0080]

[0081] in , , , , , ; For each coefficient combination in the formula Computes the correction of the edge image.

[0082] (4) Use nonlinear optimization to search for the various coefficient combinations mentioned above to obtain the optimal solution, and the corresponding parameters are the optimal values.

[0083] After the complete distortion correction of the image, the focal length of the image can be obtained according to the geometric information of the cylindrical target. After the distortion correction of the feature points of the two straight lines extracted from the calibration target selected above, there is still N n ( n =1,2) feature points. Since the cylindrical target may not be placed completely horizontally, the present invention establishes a new coordinate system rotated by θ degrees around the z-axis of the camera coordinate system, and rotates the obtained target image into an image completely perpendicular to the horizontal plane. According to the properties of rigid body transformation, it can be obtained:

[0084] (17)

[0085] Using the perspective projection model we can get:

[0086] (18)

[0087] k is the depth ratio of the focal length to the central axis of the cylinder, that is k=f / z 1; d is the height of the cylindrical target; x 1 is the cylindrical target point P 3,4 In the camera coordinate system x Axis position; y 1 is the cylindrical target point P 3,4 In the camera coordinate system y The position of the axis.

[0088] Using nonlinear least squares algorithm to obtain x 1, y 1, k The curve fitted after distortion correction is an ellipse and xoz The projection on the plane is a standard circle, so:

[0089] (19)

[0090] According to the properties of the fitted ellipse, we can get:

[0091] (20)

[0092] in, s is any non-zero scale, t 1 The center of the cylindrical target x Axis information and t 1=x 1 +Rx , t 2 is the center of the cylindrical target z Axis information, Rx is the radius of the cylindrical target. According to formula (20), the normalized focal length can be obtained a x 、 a y .

[0093] according to Steger The algorithm extracts the central stripes of the image plane. The obtained central stripes are used to perform ellipse fitting to obtain C The fitted ellipse is as follows Figure 5 As shown:

[0094] The line structured light plane intersects with the cylindrical target to form an elliptical cone in three-dimensional space. According to the camera model, the ellipse fitted at the center of the light stripe is substituted into formula (2) to obtain:

[0095] (twenty one)

[0096] in , assuming that the plane equation of the line structured light is: y=ax+cz+d , the intersection line of the line structured light and the cylindrical target can be obtained Qt exist xoz Projection equation on the surface.

[0097] (twenty two)

[0098] Since the elliptical cone xoz The projection line on the plane is a circle and we get:

[0099] (twenty three)

[0100] in, s is any non-zero scale. According to the properties, we can get a, c, d Therefore, the light plane equation is established in the camera coordinate system, and the coefficients of the light plane equation are:

[0101] (twenty four)

[0102] The internal parameter matrix of the camera obtained in this embodiment is , the first-order radial distortion coefficient of the lens k 1=-5e-07, the light plane equation is: y =0.2041 x +0.1041 z -130.1858. Calibration accuracy is given by [ u0 , v 0 , k 1 , f, a, c, d ] to evaluate the relative error.

[0103] The above embodiment is a preferred implementation of the present invention. In the complete calibration method, appropriately reducing the diameter of the cylindrical target helps to improve the camera calibration accuracy, but the improvement of the calibration accuracy by infinitely reducing the cylinder diameter is limited. When the ratio of the field of view to the cylinder diameter is about 4.93 (375mm / 76mm), the calibration accuracy is better.

Claims

1. A complete calibration method for a line structured light vision sensor based on a single cylindrical target, characterized in that: include: A laser transmitter, a camera, a tripod, and a three-dimensional cylindrical target, wherein the cylindrical target is in front of the laser transmitter, the three-dimensional target is within the field of view of the camera, and the camera is fixed on the tripod; Applying the complete calibration method includes the following steps: Step 1: The cylindrical target is placed in a suitable location. The camera captures the edge information of the cylindrical target and the light stripes at the intersection of the line structured light vision sensor and the cylindrical target. It is worth noting that due to the unique geometric characteristics of the cylindrical target, any position of the cylindrical target is allowed at this time. It only needs to distinguish the color of the cylindrical target from the external environment. Step 2: By ensuring that the modulus of the outer product of collinear vectors is zero and the distance constraint of the cylindrical symmetry is used, the camera distortion parameters are solved, and then the camera focal length is solved using the depth information of the cylindrical target edge. After obtaining the internal parameters of the camera, the image is subjected to distortion correction. Step 3: Extract the center stripes of the light after distortion correction, solve the spatial cone equation, and use the projection properties combined with the cylindrical target information to obtain the light plane equation; In the second step, when solving the camera distortion parameters, the ideal imaging point is recorded as (u p ,v p ), the lens distortion point is recorded as (u d ,v d ), the relationship between them is as follows: Where (u0v0) is the center of distortion, and the coefficient k1 represents the first-order radial distortion of the lens; For nonlinear camera lens distortion correction, the available information of the given scene before calibration can be used to solve the camera lens distortion problem; for a vector composed of any two points on a straight line, the modulus of the outer product is zero due to collinearity; for the calibration target, the feature points of two lines can be extracted on the calibration image, and each line has N n (n=1,2) feature points; P i is the ideal imaging point, P i P i+1 From two adjacent feature points P i and P i+1 The vector formed; calculate the magnitude to solve: |P i-1 P i ×P i P i+1 | 2 =[(u pi -u pi-1 )(v pi+1 -v pi )-(u pi+1 -u pi )(v pi -v pi-1 )] 2 i=2,3,…,N n -1(2) Extract the feature points of two curves on the calibration image as calibration targets; divide each curve into two symmetrical parts, each with N m (m=1,2,3,4) feature points; according to the corresponding feature points D mj (j=1,2,…,N m ), where D mj is the space coordinate in the camera coordinate system; the distance constraint of the coordinate can be obtained: |D 1j -D 3j |=|D 2j -D 4j |,j=1,2,…,N m (3) The specific implementation steps of the parameter estimation algorithm for the camera's first-order radial distortion coefficient k1 and the lens distortion center (u0, v0) are as follows: (1) Determine the range of distortion coefficient and distortion center; let r d max is the maximum distance from the distortion center to the edge of the image, and the range of k1 is Assume that the image size is w×h, w and h are the length and width of the image respectively. The distortion center is generally located in a rectangular area of 0.1w×0.1h near the center of the image, so u0∈[0.45w,0.55w], v0∈[0.45h,0.55h]; (2) Zernike moments are used to extract sub-pixel features of the edge contour of the cylindrical target to obtain the edge sub-pixel points of the original distorted image; since structured light fringes have a certain influence on image edge extraction, the present invention combines the lens distortion model to perform quadratic curve fitting on the feature points of the two straight lines extracted from the selected calibration target on the calibration image; (3) According to the value range of the above distortion parameters k1, u0, v0, the corresponding step sizes Δk1, Δu0, Δv0 are selected respectively to obtain the distortion parameter combination (k1 p ,u0 q ,v0 r ): in p=1,2,…,N1,q=1,2,…,N2,r=1,2,…,N3,Δk1=2(1 / (r d max ) 2 ) / N1,Δu0=0.1w / N2,Δv0=0.1h / N 3； For each coefficient combination (k1 p ,u0 q ,v0 r ) Calculate the correction of the edge image; (4) Using nonlinear optimization to search for the various coefficient combinations mentioned above to obtain the optimal solution, the corresponding parameters at this time are the optimal values; After the complete distortion correction of the image, the focal length of the image can be obtained according to the geometric information of the cylindrical target; after the distortion correction of the feature points of the two straight lines extracted from the calibration target selected above, there are still N n (n=1,2) feature points; since the cylindrical target may not be placed completely horizontally, a new coordinate system is established with a rotation of θ degrees around the z-axis of the camera coordinate system, and the obtained target image is rotated to be an image completely perpendicular to the horizontal plane; according to the properties of rigid body transformation, it can be obtained: Using the perspective projection model we can get: k is the depth ratio of the focal length to the central axis of the cylinder, that is, k = f / z1; d is the height of the cylindrical target; x1 is the cylindrical target point P 3,4 The position of the x-axis in the camera coordinate system; y1 is the cylindrical target point P 3,4 The position of the y-axis in the camera coordinate system; The nonlinear least squares algorithm is used to obtain x1, y1, k. After distortion correction, the fitted curve is an ellipse and its projection on the xoz plane is a standard circle. We can get: According to the properties of the fitted ellipse, we can get: Where s is any non-zero scale, t1 is the x-axis information of the center of the cylindrical target and t1 = x1 + Rx, t2 is the z-axis information of the center of the cylindrical target, and Rx is the radius information of the cylindrical target; the normalized focal length a can be obtained according to formula (7): x 、a y .

2. The calibration method according to claim 1, wherein the method for calculating the parameters of the light plane is characterized by: In the third step, during the solution of the light plane equation, the camera coordinate system (x, y, z), the world coordinate system (x w ,y w ,z w ), then the target plane equation in the camera coordinate system is: ax+by+cz+d=0 (8) Where a, b, c, d represent the four parameters of the light plane equation; The line structured light plane intersects with the cylindrical target to form an elliptical cone in three-dimensional space. According to the camera model, the ellipse fitted to the center of the light stripe can be expressed as follows: The projection equation of the intersection line of the line structured light and the cylindrical target on the xoz surface can be obtained:

Images