A machine calibration method for AOI automatic optical inspection machines in the SMT industry

Abstract

The present invention relates to the field of machine calibration, and in particular to a machine calibration method for an AOI automatic optical inspection machine in the SMT industry; Technical problem: Compared with the AOI automatic optical inspection machine under the prior art, when the machine assembly error is ignored and local image splicing is directly performed, the splicing error is very large, and the greater the deviation from the center origin of the motion system, the greater the error; Technical solution: A machine calibration method for an AOI automatic optical inspection machine in the SMT industry, including a conversion matrix T from a world coordinate system to a basic coordinate system of a motion system bw and the transformation matrix T between the camera coordinate system and the gripper coordinate system gc The method is based on the special assembly structure of the AOI automatic optical inspection machine. The images are spliced ​​through the camera calibration method, the motion system and the visual system hand-eye calibration method and the graphics processing method of the calibration image to achieve the purpose of correcting the assembly error, thereby correcting the assembly error generated during the machine assembly process.

Description

Technical Field

[0001] The present invention relates to the field of machine calibration, and in particular to a machine calibration method for an AOI automatic optical inspection machine in the SMT industry. Background Art

[0002] AOI automatic optical inspection machine is an essential inspection equipment on the SMT production line and an important guarantee for improving the quality of mounting. AOI automatic optical inspection machine is divided into 2DAOI and 3DAOI. In any case, the motion system needs to drag the visual system to move and take pictures of the PCB board on the guide rail, and splice the local images at different positions to form a complete view of the whole board. In order to ensure the splicing effect, high requirements are placed on the installation accuracy of the motion system. There are mainly three points: first, the X-axis and Y-axis of the motion system must be strictly perpendicular; second, the XOY plane of the motion system needs to be parallel to the plane where the guide rail is located; third, the motion straightness of the X-axis and Y-axis of the motion system. In the actual machine assembly process, the above three requirements are difficult to guarantee. When the machine assembly error is ignored and the local image splicing is performed directly, the splicing error is large. The greater the deviation from the center origin of the motion system, the greater the error; the assembly error comes from three aspects: 1. The assembly error caused by the non-strict perpendicularity of the X-axis and Y-axis of the motion system; 2. The assembly error caused by the non-strict parallelism of the XOY plane of the motion system and the plane where the guide rail is located; 3. The error caused by insufficient straightness of the X-axis and Y-axis of the motion system. Summary of the invention

[0003] Compared with the existing AOI automatic optical inspection machine, when the machine assembly error is ignored and local image stitching is performed directly, the stitching error is very large. The greater the deviation from the center origin of the motion system, the greater the error.

[0004] The technical solution of the present invention is: a machine calibration method for AOI automatic optical inspection machine in SMT industry:

[0005] S11: According to the travel range of the AOI machine and the size of the target to be detected, select a calibration plate of appropriate size. The calibration plate is divided into a chessboard and a circle;

[0006] S12: Make a reference line on the calibration plate, using precise photolithography or laser etching technology to make a reference line with precise length and accurate angle;

[0007] S13: making calibration points on the calibration plate, and making a plurality of calibration points of different sizes according to the standard of the relationship between pixels and physical sizes;

[0008] S14: Calibrate the accuracy of the calibration plate and verify the accuracy of the reference line and calibration points on the calibration plate;

[0009] S15: Select a camera and a bi-telecentric lens and determine the initial parameters of the visual system. Accurately define the initial parameters of the visual system according to the technical parameters of the selected camera and bi-telecentric lens, including:

[0010] (W p , H p ): Camera imaging resolution, unit: pixels;

[0011] (W mm , H mm ): visual imaging FOV size, unit: mm;

[0012] S: pixel size, in mm;

[0013] β: bi-telecentric lens magnification;

[0014] O mm : Object distance of bi-telecentric lens.

[0015] The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 1 is characterized in that: a camera photographs fixed points of a calibration plate, extracts a series of image coordinates, and estimates the resolution (W) of the complete view of the calibration plate according to the physical size of the calibration plate. f , H f ), the specific steps are as follows:

[0016] S21: Start the machine's motion system to drag the camera to move, where the motor's X-axis moves a distance of X m , the Y-axis movement distance of the motor is Y m ;

[0017] S22: Start the camera to take pictures of the fixed points on the calibration plate to obtain a series of images I(X m , Y m );

[0018] S23: Select the absolute coordinates, take I(x 0 ,y 0 );

[0019] S24: Calculate the resolution of the full view of the calibration plate according to the width and height of the calibration plate (W f , H f ), where the width of the calibration plate is P w (Unit: mm), the height of the calibration plate is P h (unit: mm), the resolution of the entire visual system is expressed as q, and the calculation formula is as follows:

[0020]

[0021] According to each local image I(X m , Y m ) Calculate H om , and use the Hough transform image processing method to extract the center or corner point. The specific steps are as follows:

[0022] S31: According to each local image I(X m , Y m ), calculate H om , the calculation formula is as follows:

[0023] t co =(-X 0 , -Y 0 , O mm ) T ;

[0024]

[0025] t cm =(-X m , -Y m , O mm ) T ;

[0026]

[0027] t q =(W f -W p ,H f -H p ,0) T ;

[0028]

[0029] If λ=1, then

[0030] S32: According to each local image I(X m , Y m ), the Hough transform image processing method is used to extract the center or corner points, and the extracted point set is recorded as:

[0031]

[0032] S33: The transformation matrix H calculated by step S31 om Transform the point set obtained in step S32 to obtain a new point set, which is recorded as:

[0033] in

[0034] The calibration plate grid structure is constructed by referring to the graphics method. The specific steps are as follows:

[0035] S41: Record the point cloud based on the new point set calculated by S33

[0036]

[0037] S42: Construct a calibration plate grid structure with reference to the graphics method, and record the grid structure as:

[0038] G={(V i ,X w,i ,N i )|i∈N G}, where Vi represents the image coordinates of mesh vertex i, X w,i Represents the world coordinates of mesh vertex i. The world coordinate system can be selected arbitrarily. N i Represents the domain index of vertex i.

[0039] According to the Laplace editing technique, the mesh G and the point cloud C are 0 Align, and obtain each local image I(X m , Y m ) The specific steps are as follows:

[0040] S51: Using Laplace editing technology to transform the mesh G and point cloud C 0 Alignment;

[0041] S52: Mesh G and point cloud C according to step S51 0 The alignment result can be obtained by obtaining each local image I(X m , Y m ) The point cloud extracted is The world coordinates of each point in are recorded as:

[0042] in Indicate point The world coordinates of .

[0043] According to each local image I(X m , Y m ), calculate the camera internal parameter matrix and the external parameter matrix of each view in step S22, and then calculate the conversion matrix from the world coordinates to the basic coordinate system of the motion system and the conversion matrix from the camera coordinate system to the gripper coordinate system according to the calculated internal parameter matrix and external parameter matrix. The specific steps are as follows:

[0044] S61: According to and The camera intrinsic parameter matrix estimation algorithm based on Zhang Zhengyou's calibration method calculates the camera intrinsic parameter matrix and the extrinsic parameter matrix of each view in step S22;

[0045] S62: Record image I(X m , Y m ) The external parameter matrix corresponding to the viewing angle is Let image I(X m , Y m ) The corresponding gripper system posture is

[0046] S63: Based on the results obtained in steps S61-S62 and The hand-eye calibration algorithm is used to calculate the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw ;

[0047] S64: Based on the results obtained in steps S61-S62 and The hand-eye calibration algorithm is used to calculate the transformation matrix T between the camera coordinate system and the gripper coordinate system. gc .

[0048] According to the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw and the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , stitching the images to achieve the purpose of correcting the assembly error, including the following steps:

[0049] S71: Create a blank image I f , whose size is (Wf,Hf);

[0050] S72: According to the formula and Calculate W f and H f ;

[0051] S73: According to (x in step S23 0 ,y 0 ),remember

[0052]

[0053] S74: Calculate T cow and T cw , the calculation formula is as follows:

[0054]

[0055] Where T gc and T bw is calculated in steps S61-S64;

[0056] S75: Calculate T om , the calculation formula is as follows:

[0057]

[0058] S76: Take in Where T*(3,4) represents the value of the 3rd row and 4th column of the matrix T*. The calculation formula for Hm is as follows:

[0059] Where K is defined as T cw The definition is n d The definition is Q is defined as

[0060] S77: The local image I(X m , Y m ) after transformation H m Transform to blank image I f , that is, one image stitching is completed;

[0061] S78: For all m∈N m Repeat steps S73-S77 to complete the stitching of all local images.

[0062] Beneficial effects of the present invention:

[0063] 1. Compared with the AOI automatic optical inspection machine under the existing technology, when the machine assembly error is ignored and the local image stitching is directly performed, the stitching error is very large. The greater the deviation from the center origin of the motion system, the greater the error. This method is based on the special assembly structure of the AOI automatic optical inspection machine. Through the camera calibration method, the motion system and the visual system hand-eye calibration method and the calibration image graphics processing method, the image is stitched to achieve the purpose of correcting the assembly error, thereby correcting the assembly error generated during the machine assembly process;

[0064] 2. Use the camera to capture the fixed points of the calibration plate, extract the coordinates of the series of images, and estimate the resolution of the complete view of the calibration plate based on the physical size of the calibration plate (W f , H f ), according to each local image I(X m , Y m ) Calculate H om The center or corner point is extracted by Hough transform image processing method, and the obtained point set is transformed to obtain a new point set. The calibration plate grid structure is constructed according to the graphics method, and the grid G ​​and point cloud C are transformed according to the Laplace editing technology. 0Align, and obtain each local image I(X m , Y m ) proposed point cloud, according to each local image I(X m , Y m ) to extract the point cloud, calculate the camera intrinsic parameter matrix and the extrinsic parameter matrix of each view, and then calculate the transformation matrix from the world coordinate system to the basic coordinate system of the motion system and the transformation matrix between the camera coordinate system and the gripper coordinate system according to the calculated intrinsic parameter matrix and extrinsic parameter matrix. bw and the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , the images are stitched together to achieve the purpose of correcting the assembly error.

[0065] 3. According to the travel range of the AOI machine and the target size to be detected, select a calibration plate of appropriate size, make a reference line on the calibration plate, use precise photolithography or laser etching technology to make a reference line with precise length and accurate angle, make calibration points on the calibration plate, make multiple calibration points of different sizes according to the standard of the relationship between pixels and physical size, calibrate the accuracy of the calibration plate, verify the accuracy of the reference line and calibration points on the calibration plate, select the camera and bi-telecentric lens and determine the initial parameters of the visual system, accurately define the initial parameters of the visual system according to the technical parameters of the selected camera and bi-telecentric lens, shoot the fixed points of the calibration plate with the camera, extract the coordinates of the series of images, and estimate the resolution of the complete view of the calibration plate according to the physical size of the calibration plate (W f , H f ), thereby reducing the error caused by insufficient straightness of the X-axis and Y-axis of the motion system. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 The diagram shows the process structure of the machine calibration method for the AOI automatic optical inspection machine in the SMT industry of the present invention;

[0067] Figure 2 The diagram shows the process structure of the image stitching of the machine calibration method for the AOI automatic optical inspection machine in the SMT industry of the present invention;

[0068] Figure 3 What is shown is a splicing effect diagram before the implementation of the machine calibration method for the AOI automatic optical inspection machine in the SMT industry of the present invention;

[0069] Figure 4 What is shown is a splicing effect diagram after the machine calibration method of the present invention is implemented for the AOI automatic optical inspection machine in the SMT industry. DETAILED DESCRIPTION

[0070] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0071] See also Figure 1-2 The present invention provides an embodiment: a machine calibration method for an AOI automatic optical inspection machine in the SMT industry:

[0072] S11: According to the travel range of the AOI machine and the size of the target to be detected, select a calibration plate of appropriate size. The calibration plate is divided into a chessboard and a circle;

[0073] S12: Make a reference line on the calibration plate, using precise photolithography or laser etching technology to make a reference line with precise length and accurate angle;

[0074] S13: making calibration points on the calibration plate, and making a plurality of calibration points of different sizes according to the standard of the relationship between pixels and physical sizes;

[0075] S14: Calibrate the accuracy of the calibration plate and verify the accuracy of the reference line and calibration points on the calibration plate;

[0076] S15: Select a camera and a bi-telecentric lens and determine the initial parameters of the visual system. Accurately define the initial parameters of the visual system according to the technical parameters of the selected camera and bi-telecentric lens, including:

[0077] (W p , H p ): Camera imaging resolution, unit: pixels;

[0078] (W mm , H mm ): visual imaging FOV size, unit: mm;

[0079] S: pixel size, in mm;

[0080] β: bi-telecentric lens magnification;

[0081] O mm : Object distance of bi-telecentric lens.

[0082] The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 1 is characterized in that: a camera photographs fixed points of a calibration plate, extracts a series of image coordinates, and estimates the resolution (W) of the complete view of the calibration plate according to the physical size of the calibration plate. f , H f ), the specific steps are as follows:

[0083] S21: Start the machine's motion system to drag the camera to move, where the motor's X-axis moves a distance of X m , the Y-axis movement distance of the motor is Y m ;

[0084] S22: Start the camera to take pictures of the fixed points on the calibration plate to obtain a series of images I(X m , Y m );

[0085] S23: Select the absolute coordinates, take I(x 0 ,y 0 );

[0086] S24: Calculate the resolution of the full view of the calibration plate according to the width and height of the calibration plate (W f , H f ), where the width of the calibration plate is P w (Unit: mm), the height of the calibration plate is P h (unit: mm), the resolution of the entire visual system is expressed as q, and the calculation formula is as follows:

[0087]

[0088] According to each local image I(X m , Y m ) Calculate H om , and use the Hough transform image processing method to extract the center or corner point. The specific steps are as follows:

[0089] S31: According to each local image I(X m , Y m ), calculate H om , the calculation formula is as follows:

[0090] t co =(-X 0 , -Y 0 , O mm ) T ;

[0091]

[0092] t cm =(-X m , -Y m , O mm ) T ;

[0093]

[0094] t q =(W f -W p ,H f -H p ,0) T ;

[0095]

[0096] If λ=1, then

[0097] S32: According to each local image I(X m , Y m ), the Hough transform image processing method is used to extract the center or corner points, and the extracted point set is recorded as:

[0098]

[0099] S33: The transformation matrix H calculated by step S31 om Transform the point set obtained in step S32 to obtain a new point set, which is recorded as:

[0100] in

[0101] The calibration plate grid structure is constructed by referring to the graphics method. The specific steps are as follows:

[0102] S41: Record the point cloud based on the new point set calculated by S33

[0103]

[0104] S42: Construct a calibration plate grid structure with reference to the graphics method, and record the grid structure as:

[0105] G={(V i ,X w,i ,N i )|i∈N G}, where Vi represents the image coordinates of mesh vertex i, X w,i Represents the world coordinates of mesh vertex i. The world coordinate system can be selected arbitrarily. N i Represents the domain index of vertex i.

[0106] According to the Laplace editing technique, the mesh G and the point cloud C are 0 Align, and obtain each local image I(X m , Y m ) The specific steps are as follows:

[0107] S51: Using Laplace editing technology to transform the mesh G and point cloud C 0 Alignment;

[0108] S52: Mesh G and point cloud C according to step S51 0 The alignment result can be obtained by obtaining each local image I(Xm , Y m ) The point cloud extracted is The world coordinates of each point in are recorded as:

[0109] in Indicate point The world coordinates of .

[0110] According to each local image I(X m , Y m ), calculate the camera internal parameter matrix and the external parameter matrix of each view in step S22, and then calculate the conversion matrix from the world coordinates to the basic coordinate system of the motion system and the conversion matrix from the camera coordinate system to the gripper coordinate system according to the calculated internal parameter matrix and external parameter matrix. The specific steps are as follows:

[0111] S61: According to and The camera intrinsic parameter matrix estimation algorithm based on Zhang Zhengyou's calibration method calculates the camera intrinsic parameter matrix and the extrinsic parameter matrix of each view in step S22;

[0112] S62: Record image I(X m , Y m ) The external parameter matrix corresponding to the viewing angle is Let image I(X m , Y m ) The corresponding gripper system posture is

[0113] S63: Based on the results obtained in steps S61-S62 and The hand-eye calibration algorithm is used to calculate the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw ;

[0114] S64: Based on the results obtained in steps S61-S62 and The hand-eye calibration algorithm is used to calculate the transformation matrix T between the camera coordinate system and the gripper coordinate system. gc .

[0115] According to the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw and the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , stitching the images to achieve the purpose of correcting the assembly error, including the following steps:

[0116] S71: Create a blank image I f , whose size is (Wf,Hf);

[0117] S72: According to the formula and Calculate W f and H f ;

[0118] S73: According to (x in step S23 0 ,y 0 ),remember

[0119]

[0120] S74: Calculate T cow and T cw , the calculation formula is as follows:

[0121]

[0122] Where T gc and T bw is calculated in steps S61-S64;

[0123] S75: Calculate T om , the calculation formula is as follows:

[0124]

[0125] S76: Take in Where T*(3,4) represents the matrix T * Calculate Hm by looking at the value in the 3rd row and 4th column. The calculation formula is as follows:

[0126] Where K is defined as T cw The definition is n d The definition is Q is defined as

[0127] S77: The local image I(X m , Y m ) after transformation H m Transform to blank image I f , that is, one image stitching is completed;

[0128] S78: For all m∈N m Repeat steps S73-S77 to complete the stitching of all local images.

[0129] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of those skilled in the art without departing from the spirit of the present invention.

Claims

1. A machine calibration method for AOI automatic optical inspection machines in the SMT industry: S11: According to the travel range of the AOI machine and the size of the target to be detected, select a calibration plate of appropriate size. The calibration plate is divided into a chessboard and a circle; S12: Make a reference line on the calibration plate, using precise photolithography or laser etching technology to make a reference line with precise length and accurate angle; S13: making calibration points on the calibration plate, and making a plurality of calibration points of different sizes according to the standard of the relationship between pixels and physical sizes; S14: Calibrate the accuracy of the calibration plate and verify the accuracy of the reference line and calibration points on the calibration plate; S15: Select a camera and a bi-telecentric lens and determine the initial parameters of the visual system. Accurately define the initial parameters of the visual system according to the technical parameters of the selected camera and bi-telecentric lens, including: (W p , H p ): Camera imaging resolution, unit: pixels; (W mm , H mm ): visual imaging FOV size, unit: mm; S: pixel size, in mm; β: bi-telecentric lens magnification; O mm : Bi-telecentric lens object distance; S2: The camera captures the fixed points of the calibration plate, extracts the coordinates of the series of images, and estimates the resolution of the full view of the calibration plate based on the physical size of the calibration plate (W f , H f ), the specific steps are as follows: S21: Start the machine's motion system to drag the camera to move, where the motor's X-axis moves a distance of X m , the Y-axis movement distance of the motor is Y m ; S22: Start the camera to take pictures of the fixed points on the calibration plate to obtain a series of images I(X m , Y m ); S23: Select the absolute coordinates, take I(x0,y0); S24: Calculate the resolution of the full view of the calibration plate according to the width and height of the calibration plate (W f , H f ), where the width of the calibration plate is P w (Unit: mm), the height of the calibration plate is P h (unit: mm), the resolution of the entire visual system is expressed as q, and the calculation formula is as follows: S3: Based on each local image I(X m , Y m ) Calculate H om , and use the Hough transform image processing method to extract the center or corner point. The specific steps are as follows: S31: According to each local image I(X m , Y m ), calculate H om , the calculation formula is as follows: t co =(-X0,-Y0,O mm ) T ; t cm =(-X m， -Y m， O mm ) T ; t q =(W f -W p ,H f -H p ,0) T ; If λ=1, then S32: According to each local image I(X m , Y m ), the Hough transform image processing method is used to extract the center or corner points, and the extracted point set is recorded as: S33: The transformation matrix H calculated by step S31 om Transform the point set obtained in step S32 to obtain a new point set, which is recorded as: in S4: Construct the calibration plate grid structure with reference to the graphics method; S5: According to the Laplace editing technique, the mesh G and the point cloud C are 0 Align, and obtain each local image I(X m , Y m )The proposed point cloud; S6: According to each local image I(X m , Y m ) the point cloud extracted, calculate the camera intrinsic parameter matrix and the extrinsic parameter matrix of each viewing angle in step S22, and then calculate the conversion matrix from the world coordinates to the basic coordinate system of the motion system and the conversion matrix from the camera coordinate system to the gripper coordinate system according to the calculated intrinsic parameter matrix and extrinsic parameter matrix; S7: Transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw And the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , the images are stitched together to achieve the purpose of correcting the assembly error.

2. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 1, characterized in that: The specific steps of S4 are as follows: S41: Record the point cloud based on the new point set calculated by S33 S42: Construct a calibration plate grid structure with reference to the graphics method, and record the grid structure as: G={(V i ,X w,i ,N i )|i∈N G }, where V i represents the image coordinates of mesh vertex i, X w,i Represents the world coordinates of mesh vertex i. The world coordinate system can be selected arbitrarily. N i Represents the domain index of vertex i.

3. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 1, characterized in that: The specific steps of S5 are as follows: S51: Using Laplace editing technology to transform the mesh G and point cloud C 0 Alignment; S52: Mesh G and point cloud C according to step S51 0 The alignment result can be obtained by obtaining each local image I(X m , Y m ) The point cloud extracted is The world coordinates of each point in are recorded as: in Indicate point The world coordinates of .

4. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 1, characterized in that: S6 specific steps are as follows: S61: According to and The camera intrinsic parameter matrix estimation algorithm based on Zhang Zhengyou's calibration method calculates the camera intrinsic parameter matrix and the extrinsic parameter matrix of each view in step S22; S62: Record image I(X m , Y m ) The external parameter matrix corresponding to the viewing angle is Let image I(X m , Y m ) The corresponding gripper system posture is S63: Based on the results obtained in steps S61-S62 The hand-eye calibration algorithm is used to calculate the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw ; S64: Based on the results obtained in steps S61-S62 The hand-eye calibration algorithm is used to calculate the transformation matrix T between the camera coordinate system and the gripper coordinate system. gc .

5. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 4, characterized in that: S7 includes the following steps: S71: Create a blank image I f , whose size is (W f ,H f ); S72: According to the formula and Calculate W f and H f : According to the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw And the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , stitching the images to achieve the purpose of correcting the assembly error, and also includes the following steps: S73: According to (x0, y0) in step S23, record S74: Calculate T cow and T cw , the calculation formula is as follows: Where T gc and T bw Calculated in steps S61-S64.

6. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 5, characterized in that: According to the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw And the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , stitching the images to achieve the purpose of correcting the assembly error, and also includes the following steps: S75: Calculate T om , the calculation formula is as follows: S76: Take in Where T * (3,4) represents the matrix T * The value of row 3 and column 4, calculate H m , the calculation formula is as follows: Where K is defined as T cw The definition is n d The definition is Q is defined as 7. The machine calibration method for an AOI automatic optical inspection machine in the SMT industry according to claim 6, characterized in that: According to the transformation matrix T from the world coordinate system to the basic coordinate system of the motion system bw And the transformation matrix T between the camera coordinate system and the gripper coordinate system gc , stitching the images to achieve the purpose of correcting the assembly error, and also includes the following steps: S77: The local image I(X m , Y m ) after transformation H m Transform to blank image I f , that is, one image stitching is completed; S78: For all m∈N m Repeat steps S73-S77 to complete the stitching of all local images.

Images