Affine template-based three-axis right-angle user coordinate system calibration method

Abstract

The invention relates to a three-axis right-angle user coordinate system calibration method based on an affine template. The method comprises the steps of Z direction calibration through laser height measurement, X and Y direction calibration through affine template matching, geometric modeling and matrix derivation algorithm, coordinate system conversion and deviation correction. According to the method, on the premise that the original structure of the system is kept unchanged, height information is obtained through the laser height measurement sensor, and coordinate system calibration and height compensation in the Z direction are achieved; on the basis of plane fitting and height measurement compensation, calibration marks are automatically recognized through a visual system, and calibration and deviation correction of a user coordinate system in the X direction and the Y direction are completed; the tool end deviation compensation amount is obtained in one-time calculation, conversion between a user coordinate system and a tool coordinate system is achieved, repeated work needing repeated calibration in a traditional coordinate system calibration method is eliminated, and the calculation efficiency and stability of a calibration algorithm are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of program-controlled manipulators, and in particular to a three-axis rectangular user coordinate system calibration method based on an affine template. Background Art

[0002] As the demand for production precision continues to increase in industries such as electronic assembly, automated manufacturing, and semiconductors, three-axis Cartesian robots are widely used in positioning and processing. To meet the needs of high-precision processing, establishing a user coordinate system (such as the origin and direction on the detection object) corresponding to the base coordinate system becomes a key step. However, traditional coordinate system calibration methods usually rely on manual alignment or mechanical adjustment, which is time-consuming and labor-intensive and difficult to achieve high precision and automation. While some existing vision- or laser-assisted calibration solutions can improve the level of automation, they often require complex calibration mechanisms or binocular depth cameras, which require high computing resources and suffer from low calibration efficiency and poor adaptability.

[0003] Chinese patent CN111445536B discloses a calibration device and method for a 3D camera. The patent sets a calibration plate with a detachable groove structure, uses a line structured light laser to perform triangulation to measure the groove cross-sectional profile, and extracts its geometric features to solve the coordinate transformation matrix. This technical solution can reduce the influence of camera lens distortion and improve calibration accuracy. However, this method has high requirements on the manufacturing process of the calibration plate, and requires additional translation stage equipment to achieve relative movement between the calibration plate and the camera, which increases the complexity and cost of the system. In addition, this solution is mainly aimed at the calibration of 3D cameras, and does not involve the automated calibration of the three-axis rectangular user coordinate system. In particular, it lacks the overall correction capability for deviations in the three directions of XYZ, and cannot be directly applied to the efficient calibration of the three-axis robot user coordinate system.

[0004] Chinese patent CN110068273B discloses a laser fusion calibration method based on a 3D model. The patent installs a laser sensor on a motion mechanism, collects grid data during the motion process, and aligns and compares it with the 3D model of the product to determine the laser posture. This method reduces the dependence on the calibration block, reduces the production cost and the time to adjust the parameters. However, this technical solution requires the acquisition of an accurate 3D model of the product to be measured in advance, and has high requirements for the trigger spacing and laser point spacing of the motion mechanism, which may lead to error accumulation during the data supplementation process. In addition, this method is mainly suitable for the calibration of static objects, and does not consider the real-time calibration requirements of the three-axis robot in the dynamic motion scene. It cannot effectively solve the problem of deviation compensation of the user coordinate system in the three directions of XYZ, which limits its application scope in automated production.

[0005] Therefore, it is necessary to design a new calibration method to solve the above problems. Summary of the Invention

[0006] The purpose of the present invention is to provide a three-axis rectangular user coordinate system calibration and correction algorithm based on affine template matching laser altimetry, which realizes efficient and accurate calibration of the user coordinate system in the XYZ directions by combining visual template matching and laser altimetry.

[0007] The present invention achieves the above-mentioned object through the following technical solution: a three-axis rectangular user coordinate system calibration method based on an affine template, wherein the device to be calibrated includes a control system, a three-axis mobile device, and an image acquisition mechanism, a laser altimeter sensor, and an actuator provided at the mobile end of the three-axis mobile device, and the steps include: S1. Laser altimeter calibration in the Z direction: Scan the user plane using the laser altimeter sensor. The laser altimeter sensor acquires a set of spatial points. The image acquisition mechanism obtains a detection image. The Z coordinate system is established and corrected by fitting the plane and calculating the angle between the normal vector and the reference axis. S2. Affine template matching calibration in X and Y directions: The control system extracts the static template features of the static template image, constructs an affine deformation model allowed for the static template in two-dimensional space, obtains an affine template, places the detection object corresponding to the static template on the user plane with the surface to be matched facing upward, uses the image acquisition mechanism to scan and obtain a detection image, slides the affine template in the detection image in an affine space, and uses an optimization algorithm to find the optimal affine parameter group so that the matching degree between the transformed template image and the actual Mark point area in the detection image is maximized, and obtains the affine linear transformation matrix, mapping translation vector, scaling factor, and surface inclination angle of the detection object; S3. Geometric modeling and matrix derivation algorithm: A base coordinate system is established with the projection of the reference point on the base of the three-axis motion device on the horizontal plane as the origin. A user coordinate system is established with the projection point of the mark point of the teaching object on the user plane as the origin. A teaching point N different from its mark point is taken on the teaching object, and the end center of the actuator contacts the teaching point. A tool coordinate system is established with the projection point M of the center point of the three-axis motion device and the actuator fixing flange on the horizontal plane as the origin. The tool end deviation compensation amount of the tool coordinate system is expressed as a matrix [ΔX, ΔY]. The teaching object is placed on the user plane in a first posture. At this time, the projection point of the Mark point of the teaching object on the user plane is the origin of the first user coordinate system. The known transformation matrix of the first user coordinate system relative to the base coordinate system is T1. Then the teaching object is placed on the user plane in a second posture. At this time, the projection point of the Mark point of the teaching object on the user plane is the origin of the second user coordinate system. The known transformation matrix of the second user coordinate system relative to the base coordinate system is T2. The three-axis moving device is based on T2×T1. -1 After the movement, the end center of the actuator is corrected to the position of the teaching point. The horizontal compensation amount during the correction is [dx, dy], and [ΔX, ΔY] is calculated by the following formula: ; S4. Coordinate system conversion and correction: The calculated tool end deviation compensation [ΔX, ΔY] is applied to the conversion of other user coordinate systems relative to the base coordinate system.

[0008] Specifically, the calibration method in the Z direction in step S1 is: Move the laser ranging sensor along the user plane area, record the height measurement data at each sampling point, and obtain a set of points with spatial positions: ; in, , is the sampling point position in the base coordinate system, is the height value measured by the laser altimeter sensor, and the plane equation is: ; The solution is to minimize the sum of squared errors: ; Written in matrix form of linear system, let , where each line is , is all A vector of values, , solve , rewritten into square matrix form as ,Right now: ; Solving by Cramer's law we get: ; Calculate normal vector and basis coordinates Angle between axes , user coordinate system The axis is defined as the direction normal to the fitting plane: .

[0009] Furthermore, the laser altimeter sensor moves along a rectangular track.

[0010] Furthermore, the solution method of the affine linear transformation matrix in step S2 is: Let the original pixel coordinates of a point in the affine template be , after affine transformation, it is mapped to: ; in, is the affine linear transformation matrix, For the translation vector, calculate the normalized cross-correlation matching metric function: ; in: is the width of the window area, is the height of the window area; is the original pixel coordinate on the affine template image The grayscale at is the grayscale mean of the window area on the affine image, To detect pixel coordinates in the image The grayscale at To detect the position in the image The grayscale mean of the area whose initial size is equal to the window area; when When it is closest to ±1, we get and Furthermore, the method for calculating the inclination angle of the surface of the detection object is as follows: , and get the scaling factor: ; Then the axial vector after transformation is: ; Change axial angle , , which is the surface inclination of the detection object.

[0011] Furthermore, the static template features of the image include grayscale, edge and texture.

[0012] Furthermore, the six degrees of freedom parameters of the affine deformation model include scale change, rotation, translation and slight shear transformation.

[0013] The beneficial effects of the technical solution of the present invention are: 1. While maintaining the original structure of the system, a laser altimeter sensor is used to obtain height information to achieve Z-direction coordinate system calibration and height compensation; 2. Based on plane fitting and height compensation, the visual system automatically identifies calibration marks and completes the calibration and correction of the user coordinate system in the X and Y directions; 3. The tool end deviation compensation is calculated once, realizing the conversion between the user coordinate system and the tool coordinate system, eliminating the repetitive work required for repeated calibration in the traditional coordinate system calibration method, and improving the computational efficiency and stability of the calibration algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a relationship diagram between equipment components and various coordinate systems; Figure 2 The schematic diagram of the matching between the template image and the detection image in step S2; Figure 3 It is a position relationship diagram between each coordinate system and each point in step S3.

[0015] The numbers in the figure represent: 1-Three-axis moving device, 2-Image acquisition mechanism, 3-Laser height sensor, 4-Actuator, 5-User plane. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below with reference to specific embodiments.

[0017] Example: When implementing the present invention, it is necessary to first clarify the basic structure of the device to be calibrated. Figure 1 As shown, it includes a control system, a three-axis mobile device 1, an image acquisition mechanism 2, a laser altimeter sensor 3 and an actuator 4 (such as a rubber head, a drill bit, etc.). The image acquisition mechanism 2, the laser altimeter sensor 3 and the actuator 4 are arranged at the mobile end of the three-axis mobile device 1. The three-axis mobile device 1 is usually composed of an X-axis, a Y-axis and a Z-axis, and can accurately control the position of the mobile end. The image acquisition mechanism 2 is used to capture the image information of the mark points on the user plane 5. These mark point image information will serve as the basic data for affine template matching and be transmitted to the control system. In this embodiment, the image acquisition mechanism 2 adopts an industrial camera. The laser altimeter sensor 3 is used to obtain the height information of the user plane 5 and transmit it to the control system. The control system can process the mark point image information and the height information of the user plane 5 to accurately control the motion trajectory of the three-axis mobile device 1, thereby completing the calibration task.

[0018] The control system directly controls the motion trajectory of the three-axis motion device 1. Its motion parameters correspond to the coordinate system of the three-axis motion device 1 itself (i.e., the base coordinate system). Inspection is based on a static template (a three-dimensional design drawing of the inspection object). However, the actual inspection is of the workpiece to be processed (i.e., the physical object being inspected), which will have some structural deviation from the three-dimensional design drawing. Furthermore, the workpiece is positioned on the user plane 5 (corresponding to the XY plane of the user coordinate system). The device's reference height and angle are uncertain, and so are the reference height and angle of the user plane 5. Therefore, there are two reasons for Z-direction tilt: systematic error due to the unevenness of the user plane 5 and the installation of the device itself, and random error due to the machining flatness of the workpiece surface. Therefore, the purpose of calibration is to more accurately convert the information on the three-dimensional design drawing into machining path information, thereby meeting high-precision machining requirements. Furthermore, due to installation issues, the actuator 4 and the three-axis motion device 1 will generate systematic errors. This systematic error, referred to as tool-end deviation compensation, cannot be directly measured. Therefore, the design idea of ​​this calibration method is to compare the two transformed position coordinates before and after the teaching point on the basis of calibration in the three directions of XYZ, and use the formula to calculate the tool end deviation compensation amount, so as to guide the application of the equipment in other user coordinate systems.

[0019] The specific implementation process of a three-axis rectangular user coordinate system calibration method based on an affine template is as follows: S1. Laser height measurement calibration in Z direction: To calibrate the user coordinate system in the Z direction and estimate the inclination angle of the test object surface, an image acquisition mechanism 2 and a laser altimeter sensor 3 are installed at the end of the three-axis motion device 1. The laser altimeter sensor 3 scans the user plane 5 to obtain a set of spatial points. By fitting the plane and calculating the angle between the normal vector and the reference axis, the Z coordinate system is established and corrected. The image acquisition mechanism 2 also obtains the test image. The three-axis motion device 1 is controlled to move the laser rangefinder 3 along a rectangular trajectory within the user plane area (avoiding collinearity). Altimeter data is recorded at each sampling point, resulting in a set of points with spatial locations: ; in, , is the sampling point position in the base coordinate, is the height value returned by the laser altimeter sensor. Let the general form of the plane equation be: ; The solution is to minimize the sum of squared errors: ; It can be written in the matrix form of a linear system, let , where each line is , is all A vector of values, , solve , rewritten into square matrix form as ,Right now: ; Solving by Cramer's law we get: ; Calculating normal vectors and base coordinates Angle between axes , user coordinate system The axis is defined as the direction normal to the fitting plane:

[0020] Compare the measured height information with the desired height of the end point to determine the Z-axis offset and possible tilt of the detection object surface, which is used to correct the vertical position of the user coordinate system.

[0021] S2. Affine template matching calibration in X and Y directions:

[0022] In practical applications, the plane of the user coordinate system (i.e., user plane 5) often exhibits a certain degree of tilt (e.g., due to uneven placement platforms or misaligned clamping of the test object). This causes variations in the image projection of the same template at different positions or postures, exhibiting various distortions such as angle, scale, and shape. To ensure the flexibility and robustness of visual template matching, this invention introduces an affine transformation matching mechanism, building upon the standard template matching algorithm, to enhance its adaptability to tilted planes.

[0023] During the ROI template creation process, not only are the static template features (grayscale, edges, and texture) of the image extracted, but an affine deformation model of the static template (the three-dimensional design of the test object) in two-dimensional space is also constructed. This includes six degrees of freedom affine parameters such as scale change, rotation, translation, and slight shear transformation, which are used to represent the possible appearance of the template after projection onto an inclined plane. During the image matching stage, the static template is captured in the test image using affine space. An optimization algorithm is used to find the optimal set of affine parameters (including rotation angle, scaling factor, shear coefficient, and translation amount) to maximize the match between the transformed template image and the actual mark point area in the test image.

[0024] Assume that the original pixel coordinates of a point in the affine template are , after affine transformation, it is mapped to: ; in, is the affine linear transformation matrix, , is the mapping translation vector.

[0025] Take the rectangular range where the static template pixels in the affine image are located as the window area and calculate the normalized cross-correlation matching metric function: ;

[0026] in: is the width of the window area (the number of pixels in the x direction), is the height of the window area (the number of pixels in the y direction); is the original pixel coordinate on the affine template image The grayscale at is the grayscale mean of the window area on the affine image, To detect pixel coordinates in the image The grayscale at To detect the position in the image The grayscale mean of an area whose initial size is equal to the window area.

[0027] when When it is closest to 1 (affine image is positive color image) or -1 (affine image is negative color image), we get and When the detection image and the template image are matched, the coordinate relationship of the corresponding points is as follows: Figure 2 shown.

[0028] The corresponding solution is to obtain the scaling factor: ; Then the axial vector after transformation is: ; Change axial angle ,like , which is the surface inclination of the detected object. Through the above component calculations, the true plane posture is restored, and the true position of the mark point in the base coordinate system is deduced after affine template matching.

[0029] S3, Geometric modeling and matrix derivation algorithm: The base coordinate system is established with the horizontal projection of the reference point on the base of the three-axis motion device 1 as its origin. The tool coordinate system is established with the horizontal projection point M of the center point of the three-axis motion device 1 and the actuator 4's fixed flange as its origin. The teach object (a specific test object used as a reference for correction) is placed on the user plane 5. The user coordinate system is established with the projection point of the teach object's mark (the reference point on the teach object used as the basis for coordinate establishment, generally the center of the teach object's graphic) on the user plane 5 as its origin. The tool end offset compensation is represented by the matrix [ΔX, ΔY]. The tool end offset compensation is caused by the relative installation distance between the center of the fixed flange and the center of the actuator 4's end. Therefore, it is a constant that is independent of the teach point position. When the center of the actuator 4's end contacts the teach point, the horizontal offset N relative to M is equal to the tool end offset compensation. The following formula is used to calculate [ΔX, ΔY] in reverse.

[0030] The specific method is: put the teaching object (in Figure 3 In the figure, the object is represented by a rectangle with a width of x1 and a height of y1, and its Mark point and the teaching point are at the diagonal relationship of the rectangle) and is placed on the user plane 5 in the first posture. At this time, the projection point of the Mark point of the teaching object on the user plane 5 is the origin of the first user coordinate system. The known transformation matrix of the first user coordinate system relative to the base coordinate system is T1 (a 2×2 matrix representing the translation and rotation relationship between the two coordinate systems). A teaching point different from its Mark point is taken on the teaching object. The projection coordinates of the teaching point in the first user coordinate system are expressed as [x1, y1] (equivalent to the coordinates of the teaching object placed in the base coordinate system in a standard posture). The coordinates of the teaching point converted to the base coordinate system are ; Then, the center of the end of the actuator 4 is used to touch the teaching point. At this time, the coordinates of the origin M1 of the first tool coordinate system in the base coordinate system are: .

[0031] That is to say, in order to make the center of the end of the actuator 4 fall on the teaching point N1, the three-axis moving device 1 is actually based on The movement of the mobile terminal is performed based on the coordinates of the mobile terminal.

[0032] Place the teaching object in the second posture on the user plane 5. The projection point of the teaching object's Mark point on the user plane 5 is the origin of the second user coordinate system. The known transformation matrix of the second user coordinate system relative to the base coordinate system is T2 (another 2×2 matrix representing the translation and rotation relationship between the two coordinate systems). The teaching point is located at N2. Because the relative positions of the teaching point and the Mark point are the same on the same teaching object, the coordinates of the teaching point in the second user coordinate system are also [x1, y1]. Then the coordinates of the teaching point converted to the base coordinate system are .

[0033] However, due to the existence of tool end deviation, the end of the actuator 4 will not reach N2 directly, but will first reach a deviation position N1', and then need to be corrected before reaching N2. After correction, the position of the origin M2 of the second tool coordinate system depends on N2, and its coordinates in the base coordinate system should be: ; Because the actuator 4 is used as the reference body, its movement is actually the movement of the three-axis motion device 1. The second posture is actually equivalent to returning from the first posture to the standard posture (the inversion of T1), and then from the standard posture to the second posture (T2). Before correction, the coordinates M1' of the origin of the tool coordinate system in the base coordinate system are: M1'=T2×M1×T1 -1 ; = T2×([x1,y1] + T1 -1 × [ΔX, ΔY]) = T2×[x,y]+ T2×T1 -1 × [ΔX,ΔY]

[0034] However, when the three-axis motion device 1 moves in this way, the end of the actuator 4 will fall at position N1' instead of N2 where the teaching point is actually located, so horizontal compensation is required. Before correction, the coordinates of the center projection N1' of the end of the actuator 4 in the base coordinate system are: N1'= M1' -[ΔX,ΔY] = T2×[x,y]+ T2×T1 -1 × [ΔX,ΔY]- [ΔX,ΔY] = N2+ (T2×T1 -1 -E)×[ΔX,ΔY]; Let the horizontal compensation amount of N1' corrected to N2 (M1' corrected to M2 synchronously) be [dx,dy], then: N2-N1' = [dx,dy] = (E-T2×T1 -1 )×[ΔX,ΔY] (E is a 2×2 identity matrix); T1, T2, and [dx,dy] are all measurable values. Substituting them into the above formula can calculate [ΔX, ΔY].

[0035] In order to make the calculation result more accurate, the difference between the two postures of the teaching object can be deliberately enlarged. In this way, dx and dy will be larger in the image, making it easier to detect and distinguish.

[0036] The final formula shows that [ΔX, ΔY] is independent of [x1, y1]. This means that the tool end deviation compensation can be calculated once and applied to all machining scenarios of the system. This eliminates the need for repeated calibration of user coordinates while ensuring the efficiency and stability of the calibration algorithm.

[0037] S4. Coordinate system conversion and correction: The control system applies the tool-side deviation compensation [ΔX, ΔY] to the conversion of other user coordinate systems relative to the base coordinate system. In subsequent machining programs, all input user coordinate system commands are first converted to the base coordinate system using a transformation matrix, with the tool-side deviation compensation [ΔX, ΔY] incorporated. The three-axis motion device 1 then moves according to these adjusted coordinates. This method achieves precise mapping and real-time deviation correction between the user coordinate system and the base coordinate system, ensuring the positioning accuracy of the actuator 4 end relative to the machining object.

[0038] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the inventive concept of the present invention, which all fall within the scope of protection of the present invention.

Claims

1. A method for calibrating a three-axis rectangular user coordinate system based on an affine template. The calibration equipment includes a control system, a three-axis mobile device, an image acquisition mechanism, a laser altimeter sensor, and an actuator. The image acquisition mechanism, the laser altimeter sensor, and the actuator are arranged at the mobile end of the three-axis mobile device. The image acquisition mechanism is used to capture image information of mark points on the user plane and transmit it to the control system. The laser altimeter sensor is used to obtain height information of the user plane and transmit it to the control system. The control system processes the mark point image information and the height information of the user plane to accurately control the motion trajectory of the three-axis mobile device. It is characterized by The steps include: S1. Laser altimeter calibration in the Z direction: Scan the user plane using the laser altimeter sensor. The laser altimeter sensor acquires a set of spatial points. The image acquisition mechanism obtains a detection image. The Z coordinate system is established and corrected by fitting the plane and calculating the angle between the normal vector and the reference axis. S2. Affine template matching calibration in X and Y directions: The control system extracts the static template features of the static template image, constructs an affine deformation model allowed for the static template in two-dimensional space, obtains an affine template, places the detection object corresponding to the static template on the user plane with the surface to be matched facing upward, uses the image acquisition mechanism to scan and obtain a detection image, slides the affine template in the detection image in an affine space, and uses an optimization algorithm to find the optimal affine parameter group so that the matching degree between the transformed template image and the actual Mark point area in the detection image is maximized, and obtains the affine linear transformation matrix, mapping translation vector, scaling factor, and surface inclination angle of the detection object; S3. Geometric modeling and matrix derivation algorithm: A base coordinate system is established with the projection of the reference point on the base of the three-axis motion device on the horizontal plane as the origin. A user coordinate system is established with the projection point of the mark point of the teaching object on the user plane as the origin. A teaching point N different from its mark point is taken on the teaching object, and the end center of the actuator contacts the teaching point. A tool coordinate system is established with the projection point M of the center point of the three-axis motion device and the actuator fixing flange on the horizontal plane as the origin. The tool end deviation compensation amount of the tool coordinate system is expressed as a matrix [ΔX, ΔY]. The teaching object is placed on the user plane in a first posture. At this time, the projection point of the Mark point of the teaching object on the user plane is the origin of the first user coordinate system. The known transformation matrix of the first user coordinate system relative to the base coordinate system is T1. Then the teaching object is placed on the user plane in a second posture. At this time, the projection point of the Mark point of the teaching object on the user plane is the origin of the second user coordinate system. The known transformation matrix of the second user coordinate system relative to the base coordinate system is T2. The three-axis moving device is based on T2×T1. -1 After the movement, the end center of the actuator is corrected to the position of the teaching point. The horizontal compensation amount during the correction is [dx, dy], and [ΔX, ΔY] is calculated by the following formula: ； S4. Coordinate system conversion and deviation correction: The control system applies the calculated tool end deviation compensation [ΔX, ΔY] to the conversion of other user coordinate systems relative to the base coordinate system.

2. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 1, characterized in that: The calibration method for the Z direction in step S1 is: Move the laser ranging sensor along the user plane area, record the height measurement data at each sampling point, and obtain a set of points with spatial positions: ; in, is the sampling point position in the base coordinate system, is the height value measured by the laser altimeter sensor, and the plane equation is: ; The solution is to minimize the sum of squared errors: ; Written in matrix form of linear system, let , where each line is , , is all A vector of values, , solve , rewritten into square matrix form as ,Right now: ； Solving by Cramer's law we get: ； Calculate normal vector and basis coordinates Angle between axes , user coordinate system The axis is defined as the direction normal to the fitting plane: .

3. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 2, characterized in that: The laser altimeter sensor moves along a rectangular track.

4. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 3, characterized in that: The solution method of the affine linear transformation matrix in step S2 is: Let the original pixel coordinates of a point in the affine template be , after affine transformation, it is mapped to: ； in, is the affine linear transformation matrix, For the translation vector, calculate the normalized cross-correlation matching metric function: ； in: is the width of the window area, is the height of the window area; is the original pixel coordinate on the affine template image The grayscale at is the grayscale mean of the window area on the affine image, To detect pixel coordinates in the image The grayscale at To detect the position in the image The grayscale mean of the area whose initial size is equal to the window area; when When it is closest to ±1, we get and .

5. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 4, characterized in that: The calculation method of the inclination angle of the surface of the detection object is as follows: , and get the scaling factor: ; Then the axial vector after transformation is: ; Change axial angle , , which is the surface inclination of the detection object.

6. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 4, characterized in that: The static template features of the image include grayscale, edge and texture.

7. The three-axis rectangular user coordinate system calibration method based on an affine template according to claim 4, characterized in that: The six degrees of freedom parameters of the affine deformation model include scale change, rotation, translation and slight shear transformation.

Images