Large-size ultra-thin glass edge contour identification method

Abstract

The invention provides a large-size ultra-thin glass edge contour recognition method, which comprises the following steps of: adopting a high-precision machine vision system, firstly, photographing an ultra-thin glass workpiece for multiple times by using a camera of the machine vision system, fixing a central displacement value of the camera during each photographing, and covering a part of content of the workpiece by each obtained image; high-precision sub-pixel edge identification is carried out on each image, identified pixel point edge coordinates are converted into coordinates under a workpiece coordinate system, coordinates of all points on the edge of the workpiece are obtained after all the images forming the workpiece are processed, the centroid positions of the coordinates are calculated to serve as polar coordinate original points, and the coordinates of all the points on the edge of the workpiece are obtained; the obtained coordinates of all the points on the edge of the workpiece are arranged anticlockwise from small to large according to the polar angles of the polar coordinates, the coordinates of all the anticlockwise arranged points forming the large-size ultra-thin glass workpiece are obtained, and multi-image splicing edge recognition and positioning of the large-size ultra-thin glass workpiece are completed; according to the method, the edge contour of the large-size ultrathin glass in any shape can be identified and extracted.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision, and in particular to a method for identifying the edge profile of large-size ultra-thin glass. Background Art

[0002] Large-size ultra-thin glass generally refers to large-size glass with a thickness less than 1.5 mm, and in some advanced application scenarios, its thickness can even reach less than 0.1 mm. This ultra-thin characteristic endows the glass material with a series of unique properties, such as light weight, high transparency, high stiffness, etc., making it show broad application prospects in multiple fields such as optics, electronic devices, biosensing, and new energy. Compared with traditional thick glass, large-size ultra-thin glass faces many challenges in the production and processing process. First of all, due to its extremely thin thickness, its mechanical strength is relatively low, which makes it extremely vulnerable to damage or produce micro-deformations during processing and handling, increasing the difficulty of edge recognition. Secondly, the optical properties of ultra-thin glass are also different from ordinary glass. For example, changes in light transmittance and reflectance may cause traditional lighting and imaging methods to fail. In order to accurately capture the fine features of the ultra-thin glass edge, a high-resolution industrial camera needs to be used, combined with a ring light illumination method to provide sufficient contrast to ensure the accuracy of edge detection. Edge recognition is one of the key steps in the processing of large-size ultra-thin glass. Considering various possible defect types of ultra-thin glass, such as bubbles, stones, unevenness, scratches, etc., in order to accurately locate and measure the position and shape of the glass edge, a high-precision sub-pixel edge detection algorithm must be used. Summary of the Invention

[0003] The present invention proposes a method for identifying the edge profile of large-size ultra-thin glass, which can realize the identification and extraction of the edge profile of large-size ultra-thin glass with any shape.

[0004] The present invention adopts the following technical solutions.

[0005] A method for identifying the edge profile of large-size ultra-thin glass, the method uses a high-precision machine vision system. First, the camera of the machine vision system takes multiple photos of the ultra-thin glass workpiece, the central displacement value of the camera is fixed each time, and each obtained image covers a part of the content of the workpiece. Then, high-precision sub-pixel edge recognition is performed on each image, and the edge coordinates of the recognized pixel points are converted into coordinates in the workpiece coordinate system. After all the images of the workpiece are processed, the coordinates of all points on the workpiece edge are obtained, the centroid position is calculated, and it is used as the origin of the polar coordinate. Finally, the coordinates of all points on the workpiece edge obtained are arranged in ascending order of the polar angle of the polar coordinate in the counterclockwise direction, and the coordinates of all points arranged counterclockwise forming the large-size ultra-thin glass workpiece are obtained, completing the multi-image stitching edge recognition and positioning of the large-size ultra-thin glass workpiece.

[0006] The recognition method includes the following steps;

[0007] Step S1: Photograph and position the glass workpiece according to the workpiece coordinate system. For details, see Figure 1 ;

[0008] Step S2: The camera first moves to a certain position of the workpiece to take a photo and form an image of the workpiece;

[0009] Step S3: Perform sub-pixel edge recognition on the photo obtained by photographing. The sub-pixel recognition method is improved based on the pixel-level Canny algorithm. A parabola fitting is performed on three points (A, B, C) along the gradient direction. The calculation of the maximum offset η is shown in Equation (1.1). The values of the corrected sub-pixel coordinates in the x and y directions, e x 、e y The calculation formula is shown in Equation (1.2). In the image, the continuous actual edge of the workpiece is the longest on the outermost side. Select the longest segment recognized as the edge of the workpiece;

[0010]

[0011]

[0012] Step S4: Perform coordinate transformation to convert the edge pixel coordinates of the workpiece into coordinates in the workpiece coordinate system, and summarize all the coordinate data to obtain the coordinate data of all points on a partial edge of the workpiece;

[0013] Assume that the point in the pixel coordinate system is (u, v), the point in the workpiece coordinate system is (X, Y), the translation vector is (tx, ty), and the point p in the pixel coordinate system is [u, v, 1] T， The point p in the workpiece coordinate system is [X, Y, 1] T , in a two-dimensional plane, only a translation transformation needs to be performed on the coordinates. The formula for the transformation matrix is Equation 1.3 below. The specific formula for converting pixel coordinates to workpiece coordinates is shown in Equation 1.4,

[0014] For details of the schematic diagram, see Figure 4 ;

[0015]

[0016]

[0017] Step S5: When the workpiece size is large and multiple imaging is required to obtain the complete edge contour of the workpiece, the specific movement logic of the camera is as follows: Use the last point of the recognized workpiece edge as the center point of the field of view for the next camera photo imaging. Move the camera to the specified position and perform the next photo imaging process, and continue to execute Step S2 for edge recognition;

[0018] Repeat the steps of step S3-step S4-step S2 to perform the operations of photographing, edge recognition, and coordinate sorting until the camera captures a complete image of the workpiece edge contour. Figure 2 ;

[0019] Step S6: Summarize the coordinates of all points of the workpiece contour in the workpiece coordinate system, store them in a point set, and calculate the centroid coordinates of the workpiece contour according to the coordinates of all points constituting the workpiece edge contour;

[0020] Assume that the point set P = {P1(x1,y1),P2(x2,y2),...,Pn(xn,yn)}P = {P1(x1,y1),P2(x2,y2),...,Pn(xn,yn)} is the set of closed edge contours that constitute the workpiece, and its centroid can be calculated by the following formula, where A is the area of the closed contour, (C x C y ) are the centroid coordinates of the contour;

[0021]

[0022]

[0023]

[0024] Step S7: The centroid coordinates (C x C y ) is set as the origin of the polar coordinate system (see Figure 5 ), converting the workpiece coordinate system coordinates of the workpiece edge pixel points into polar coordinate system coordinates;

[0025] Step S8: Sort in counterclockwise order from small to large polar angles; the steps of polar coordinate sorting are as follows: take the centroid coordinates of the edge contour of the workpiece obtained in step S7 as the reference point, and for each coordinate point identified by the edge on each image, calculate the angle and distance of each coordinate point relative to the reference point. The calculation formula is as follows: i = atan2(P iy -C y ,P ix -C x ) Formula 1.8;

[0026] where θ i is the corresponding polar angle, Pi x 、Pi y is the coordinate of the point, C x , C yis the centroid coordinate of the workpiece edge contour obtained in step S7; all points are sorted in a counterclockwise direction from small to large according to the calculated angle; step S9: the coordinates of all points are summarized to obtain the coordinate set of the accurate edge points of the large-size ultra-thin glass workpiece finally identified by sub-pixel edge recognition, without splicing the images, and the edge of the glass workpiece drawn according to the coordinates of all edge points is shown in FIG. Figure 5 .

[0027] In step S1, a high-resolution camera is used in combination with a low-illuminance blue ring light source to take a photo.

[0028] In step S1, the workpiece coordinate system is defined as follows: the origin of the workpiece coordinate system is determined by the servo driver that controls the precise movement of the three axes in the CNC machine tool. Specifically, when each axis on the machine tool reaches a suitable position, the servo driver's return to zero method 35 is used to set the current position as the zero point of the machine tool's workpiece absolute coordinate system. Thereafter, the movement of each axis on the machine tool uses this point as a reference point.

[0029] The ultra-thin glass workpiece is fixed by vacuum adsorption by a fixture. In step S2, when the camera moves to a certain position of the workpiece for the first time to take a picture of the workpiece, the method for selecting the position is related to the fixture for vacuum adsorption of the glass, specifically: in order to ensure the stability, safety and surface damage of the glass workpiece installation, the glass size does not exceed the clamping range of the vacuum adsorption of the fixture, the installation position of the fixture is known in the workpiece coordinate system, and when the clamping range of the fixture is known, the first position of the camera's photographing center selects one of the four vertex positions in the clamping range of the fixture. The positions of the four vertices are also known in the workpiece coordinate system. By calculating the distance between the positions of the four vertices of the fixture and the current position of the camera, the point closest to the camera is selected as the camera center position for taking the first photo, and the camera is moved to this point to take the first photo.

[0030] In step S3, the image edge effect after recognition is detailed in Figure 3 , including the identified multiple edge lines.

[0031] In step S8, the polar angles are sorted counterclockwise starting from 0°.

[0032] In step S9, the coordinate set of the accurate edge points of the ultra-thin glass workpiece is arranged in a counterclockwise direction.

[0033] In step S9, after obtaining the coordinate set of the accurate edge points of the ultra-thin glass workpiece, the coordinate set is arranged in reverse order, and processing path information is generated according to the coordinate set.

[0034] The machine vision system includes a camera and a lens disposed directly above the ultra-thin glass workpiece. The ultra-thin glass workpiece is fixed by a fixture. When taking pictures, an LED light source located above the side of the ultra-thin glass workpiece illuminates the ultra-thin glass workpiece.

[0035] The present invention designs and implements a registration and recognition method for the edge contour of large-size ultra-thin glass. This method takes pictures of the ultra-thin glass through a machine vision system. Since the size of the ultra-thin glass is large and the field of view of the camera is limited, in order to obtain the edge recognition coordinates of the workpiece with high precision, a small picture of the workpiece is taken each time a picture is taken. Edge recognition processing at the sub-pixel accuracy level is performed on each image of the workpiece. The coordinates of the edge pixel points recognized are converted to the workpiece coordinate system where the workpiece is located. The coordinates of the edge pixel points in all images are converted. When the coordinates of all edge points of the workpiece are known, the centroid position of the workpiece is calculated and selected as the origin of the polar coordinates. The coordinates of all the edge pixel points obtained by shooting are rearranged in ascending order of the polar angle with this point as the origin of the polar coordinates, and the coordinates of all the points forming the edge contour of the ultra-thin glass arranged in counterclockwise order are obtained, which can realize the recognition and extraction of the edge contour of large-size ultra-thin glass of any shape. Brief Description of the Drawings

[0036] The following further details the present invention in conjunction with the drawings and specific embodiments:

[0037] Attached Figure 1 is a schematic diagram of the photographing mechanism in step S1;

[0038] Attached Figure 2 is a schematic diagram of taking pictures of the workpiece with the camera's field of view and movement;

[0039] Attached Figure 3 is a schematic diagram of the high-precision sub-pixel recognition effect;

[0040] Attached Figure 4 is a schematic diagram of the camera coordinate system and the workpiece coordinate system;

[0041] Attached Figure 5 is a schematic diagram of the setting of the origin of the polar coordinate sorting and the complete counterclockwise edge path after the edge contour is spliced;

[0042] Attached Figure 6 is a schematic diagram of the process of the edge contour recognition method for large-size ultra-thin glass. Specific Embodiments

[0043] As shown in the figure, a method for identifying the edge contour of a large-sized ultra-thin glass. The method uses a high-precision machine vision system. First, the camera of the machine vision system takes multiple photos of the ultra-thin glass workpiece. Each time the photo is taken, the central displacement value of the camera is fixed, and each obtained image covers a part of the content of the workpiece. Then, high-precision sub-pixel edge recognition is performed on each image, and the edge coordinates of the identified pixel points are converted into coordinates in the workpiece coordinate system. After processing all the images that make up the workpiece, the coordinates of all points on the workpiece edge are obtained, the centroid position is calculated, and it is set as the origin of the polar coordinates. Finally, the coordinates of all points on the obtained workpiece edge are arranged in ascending order of the polar angle of the polar coordinates in the counterclockwise direction, and the coordinates of all points arranged in the counterclockwise direction that make up the large-sized ultra-thin glass workpiece are obtained, completing the multi-image stitching edge recognition and positioning of the large-sized ultra-thin glass workpiece.

[0044] The recognition method includes the following steps;

[0045] Step S1: Take pictures and position the glass workpiece according to the workpiece coordinate system, see details in Figure 1 ;

[0046] Step S2: The camera first moves to a certain position of the workpiece to take a picture of the workpiece for imaging;

[0047] Step S3: Perform sub-pixel edge recognition on the taken photo. The sub-pixel recognition method is improved based on the pixel-level Canny algorithm. Parabolic fitting is performed on three points (A, B, C) along the gradient direction. The calculation of the maximum offset η is shown in Equation (1.1). The values of the corrected sub-pixel coordinates in the x and y directions, e x 、e y The calculation formula is shown in Equation (1.2). In the image, the continuous actual edge of the workpiece is the longest on the outermost side. Select the longest segment identified as the identified workpiece edge;

[0048]

[0049]

[0050] Step S4: Perform coordinate transformation, convert the edge pixel coordinates of the workpiece into coordinates in the workpiece coordinate system, and summarize all the coordinate data to obtain the coordinate data of all points on a part of the edge of the workpiece;

[0051] Assume that the point in the pixel coordinate system is (u, v), the point in the workpiece coordinate system is (X, Y), the translation vector is (tx, ty), and the point p in the pixel coordinate system is [u, v, 1] T， The point p in the workpiece coordinate system is [X, Y, 1] T, two-dimensional plane, only the coordinates need to be translated, the transformation matrix formula is as follows 1.3, the specific formula for converting pixel coordinates to workpiece coordinates is shown in 1.4,

[0052] See the schematic diagram for details Figure 4 ;

[0053]

[0054]

[0055] Step S5: When the workpiece is large and multiple imaging is required to obtain the complete edge contour of the workpiece, the specific movement logic of the camera is: the last point of the workpiece edge identified is used as the center point of the camera's imaging field of view for the next photo, the camera is moved to the specified position and the next photo imaging process is performed, and step S2 is continued to perform edge recognition;

[0056] Repeat the steps of step S3-step S4-step S2 to perform the operations of photographing, edge recognition, and coordinate sorting until the camera captures a complete image of the workpiece edge contour. Figure 2 ;

[0057] Step S6: Summarize the coordinates of all points of the workpiece contour in the workpiece coordinate system, store them in a point set, and calculate the centroid coordinates of the workpiece contour according to the coordinates of all points constituting the workpiece edge contour;

[0058] Assume that the point set P = {P1(x1,y1),P2(x2,y2),...,Pn(xn,yn)}P = {P1(x1,y1),P2(x2,y2),...,Pn(xn,yn)} is the set of closed edge contours that constitute the workpiece, and its centroid can be calculated by the following formula, where A is the area of the closed contour, (C x C y ) are the centroid coordinates of the contour;

[0059]

[0060]

[0061]

[0062] Step S7: The centroid coordinates (C x C y ) is set as the origin of the polar coordinate system (see Figure 5 ), converting the workpiece coordinate system coordinates of the workpiece edge pixel points into polar coordinate system coordinates;

[0063] Step S8: Sort in counterclockwise order from small to large polar angles; the steps of polar coordinate sorting are as follows: take the centroid coordinates of the workpiece edge contour obtained in step S7 as the reference point, and for each coordinate point identified on the edge of each image, calculate the angle and distance of each coordinate point relative to the reference point. The calculation formula is as follows:

[0064] θ i = atan2(P iy -C y ,P ix -C x ) Formula 1.8;

[0065] Where θi is the corresponding polar angle, Pi x 、Pi y is the coordinate of the point, C x , C y is the centroid coordinate of the workpiece edge contour obtained in step S7; all points are sorted in a counterclockwise direction from small to large according to the calculated angle; step S9: the coordinates of all points are summarized to obtain the coordinate set of the accurate edge points of the large-size ultra-thin glass workpiece finally identified by sub-pixel edge recognition, without splicing the images, and the edge of the glass workpiece drawn according to the coordinates of all edge points is shown in FIG. Figure 5 .

[0066] In step S1, a high-resolution camera is used in combination with a low-illuminance blue ring light source to take a photo.

[0067] In step S1, the workpiece coordinate system is defined as follows: the origin of the workpiece coordinate system is determined by the servo driver that controls the precise movement of the three axes in the CNC machine tool. Specifically, when each axis on the machine tool reaches a suitable position, the servo driver's return to zero method 35 is used to set the current position as the zero point of the machine tool's workpiece absolute coordinate system. Thereafter, the movement of each axis on the machine tool uses this point as a reference point.

[0068] The ultra-thin glass workpiece is fixed by vacuum adsorption by a fixture. In step S2, when the camera moves to a certain position of the workpiece for the first time to take a picture of the workpiece, the method for selecting the position is related to the fixture for vacuum adsorption of the glass, specifically: in order to ensure the stability, safety and surface damage of the glass workpiece installation, the glass size does not exceed the clamping range of the vacuum adsorption of the fixture, the installation position of the fixture is known in the workpiece coordinate system, and when the clamping range of the fixture is known, the first position of the camera's photographing center selects one of the four vertex positions in the clamping range of the fixture. The positions of the four vertices are also known in the workpiece coordinate system. By calculating the distance between the positions of the four vertices of the fixture and the current position of the camera, the point closest to the camera is selected as the camera center position for taking the first photo, and the camera is moved to this point to take the first photo.

[0069] In step S3, for the edge effect of the recognized image, see Figure 3 , which includes multiple recognized edge lines.

[0070] In step S8, the polar angles are sorted counterclockwise starting from 0°.

[0071] In step S9, the coordinate set of the accurate edge points of the ultra-thin glass workpiece is arranged in the counterclockwise direction.

[0072] In step S9, after obtaining the coordinate set of the accurate edge points of the ultra-thin glass workpiece, the coordinate set is arranged in reverse order, and machining path information is generated based on the coordinate set.

[0073] The machine vision system includes a camera and a lens disposed directly above the ultra-thin glass workpiece. The ultra-thin glass workpiece is fixed by a fixture. When taking pictures, an LED light source located above the side of the ultra-thin glass workpiece is used to illuminate the ultra-thin glass workpiece.

[0074] A registration and recognition method for the edge contour of large-size ultra-thin glass proposed in this example. This method takes pictures of the ultra-thin glass through a machine vision system. Since the size of the ultra-thin glass is large and the field of view of the camera is limited, in order to obtain the coordinates of the workpiece edge with high precision, a small picture of the workpiece is taken each time. Edge recognition processing at the sub-pixel accuracy level is performed on each image of the workpiece. The coordinates of the edge pixel points recognized are converted to the workpiece coordinate system where the workpiece is located. The edge pixel point coordinates in all images are converted. When the coordinates of all edge points of the workpiece are known, the centroid position of the workpiece is calculated and selected as the polar coordinate origin. The coordinates of all the edge pixel points obtained by shooting are rearranged in ascending order of polar angle with this point as the polar coordinate origin, and the coordinates of all points forming the edge contour of the ultra-thin glass arranged in the counterclockwise order are obtained, realizing the recognition and extraction of the edge contour of large-size ultra-thin glass with any shape.

[0075] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions described in the foregoing examples, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. A method for identifying the edge profile of large-sized ultra-thin glass, characterized in that: The method uses a high-precision machine vision system. First, the camera of the machine vision system takes multiple photos of the ultra-thin glass workpiece. Each time the photo is taken, the central displacement value of the camera is fixed, and each obtained image covers a part of the workpiece. Then, high-precision sub-pixel edge recognition is performed on each image, and the edge coordinates of the recognized pixel points are converted into coordinates in the workpiece coordinate system. After processing all the images that make up the workpiece, the coordinates of all points on the workpiece edge are obtained, the centroid position is calculated, and it is set as the origin of the polar coordinates. Finally, the coordinates of all points on the obtained workpiece edge are arranged in ascending order of the polar angle of the polar coordinates in the counterclockwise direction, and the coordinates of all points arranged in the counterclockwise direction that make up the large-size ultra-thin glass workpiece are obtained, completing the multi-image stitching edge recognition and positioning of the large-size ultra-thin glass workpiece.

2. The method for identifying the edge profile of a large-size ultra-thin glass according to claim 1, characterized in that: It includes the following steps; Step S1: Take a photo and position the glass workpiece according to the workpiece coordinate system; Step S2: The camera first moves to a certain position of the workpiece to take a photo and form an image; Step S3: Perform sub-pixel edge recognition on the photo obtained by taking a picture. The sub-pixel recognition method performs parabolic fitting on three points (A, B, C) along the gradient direction. The calculation of the maximum offset η is shown in Equation 1.

1. The values of the corrected sub-pixel coordinates in the x and y directions, e x , e y The calculation formula is shown in Equation 1.2 In the image, the continuous actual edge of the workpiece is the longest on the outermost side. Select the longest segment recognized as the recognized workpiece edge; Step S4: Perform coordinate transformation to convert the edge pixel coordinates of the workpiece into coordinates in the workpiece coordinate system, and summarize all the coordinate data to obtain the coordinate data of all points on a part of the workpiece edge; Assume that the point in the pixel coordinate system is (u, v), the point in the workpiece coordinate system is (X, Y), the translation vector is (tx, ty), and the point p in the pixel coordinate system is [u, v, 1] T， The point p in the workpiece coordinate system is [X, Y, 1] T , for a two-dimensional plane, only a translation transformation needs to be performed on the coordinates. The formula for the transformation matrix is Equation 1.3 below. The specific formula for converting pixel coordinates to workpiece coordinates is shown in Equation 1.4 Step S5: When the workpiece size is large and multiple imaging is required to obtain the complete workpiece edge contour, the specific movement logic of the camera is as follows: Take the last point of the recognized workpiece edge as the center point of the next camera photo imaging field of view, move the camera to the specified position and perform the next photo imaging process, and continue to execute Step S2 for edge recognition; Repeatedly execute the operations of photo imaging, edge recognition, and coordinate sorting according to the steps of Step S3 - Step S4 - Step S2 until the camera finishes taking pictures of the complete workpiece edge contour picture; Step S6: Summarize the point coordinates in the workpiece coordinate system of all the obtained workpiece contours, store them in a total set of points, and calculate the centroid coordinates of the workpiece contour according to the coordinates of all points that make up the workpiece edge contour; Suppose the point set P = {P1(x1, y1), P2(x2, y2),..., Pn(xn, yn)} is a set of closed edge contours that make up the workpiece, and its centroid can be calculated by the following formula, where A is the area of the closed contour, (C x C y ) are the centroid coordinates of the contour; Step S7: Set the centroid coordinates (C x C y ) of the workpiece edge contour as the origin of the polar coordinate system, and convert the workpiece coordinate system coordinates of the workpiece edge pixel points into polar coordinate system coordinates; Step S8: Sort in counterclockwise order from the smallest polar angle; the steps for polar coordinate sorting are as follows: Using the centroid coordinates of the workpiece edge contour obtained in Step S7 as the reference point, for each coordinate point recognized on the edge of each image, calculate the angle and distance of each coordinate point relative to the reference point. The calculation formula is as shown in Equation θ i = atan2(P iy - C y , P ix - C x ) Equation 1.8; where θi is the corresponding polar angle, and Pi x and Pi y are the coordinates of the points, and C x and C y are the centroid coordinates of the workpiece edge contour obtained in step S7; sort all the points in ascending order in the counterclockwise direction according to the calculated angles; Step S9: Summarize the coordinates of all points to obtain the coordinate set of the accurate edge points of the large-size ultra-thin glass workpiece with sub-pixel edge recognition finally recognized.

3. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 2, characterized in that: In Step S1, a high-resolution camera is used in combination with a low-illumination blue ring light source for taking pictures.

4. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 2, characterized in that: In Step S1, the definition of the workpiece coordinate system is as follows: The origin of the workpiece coordinate system is determined by the servo driver that controls the precise movement of the three axes in the numerical control machine tool. Specifically, when each axis on the machine tool reaches the appropriate position, use the zero return method 35 of the servo driver to set the current position as the zero point of the absolute coordinate system of the workpiece on the machine tool. After that, the movement of each axis on the machine tool uses this point as the reference point.

5. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 4, characterized in that: The ultra-thin glass workpiece is adsorbed and fixed by a fixture in a vacuum manner. In step S2, when the camera first moves to a certain position of the workpiece to take a picture of the workpiece, the selection method of this position is related to the fixture for vacuum-adsorbing the glass. Specifically, in order to ensure the stability, safety and no surface damage of the glass workpiece installation, the size of the glass does not exceed the clamping range of the fixture for vacuum adsorption. The installation position of the fixture is known in the workpiece coordinate system. Given the clamping range of the fixture, the first position of the camera shooting center selects one of the four vertex positions within the clamping range of the fixture, and the positions of these four vertices are also known in the workpiece coordinate system. By calculating the distances between the positions of the four vertices of the fixture and the current position of the camera, the point closest to the camera is selected as the camera center position for taking the first picture, and the camera is moved to this point to perform the first picture taking process.

6. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 2, characterized in that: In step S3, the recognized image edge includes multiple recognized edge lines.

7. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 2, characterized in that: In step S8, the polar angle is sorted counterclockwise starting from 0°.

8. A method for identifying the edge profile of a large-size ultra-thin glass according to claim 8, characterized in that: In step S9, the coordinate set of the accurate edge points of the ultra-thin glass workpiece is arranged in counterclockwise order.

9. A method for identifying the edge contour of a large-size ultra-thin glass according to claim 2, characterized in that: In step S9, after obtaining the coordinate set of the accurate edge points of the ultra-thin glass workpiece, the coordinate set is arranged in reverse order, and the processing path information is generated according to the coordinate set.

10. A method for identifying the edge contour of a large-size ultra-thin glass according to claim 1, characterized in that: The machine vision system includes a camera and a lens disposed directly above the ultra-thin glass workpiece. The ultra-thin glass workpiece is fixed by a fixture. When taking pictures, the ultra-thin glass workpiece is illuminated by an LED light source located above the side of the ultra-thin glass workpiece.

Images