A method and apparatus for online inspection of flat glass

Abstract

This invention discloses an online inspection method and apparatus for flat glass. The method includes sensing the arrival of the glass using a photoelectric sensor, providing precise position feedback using a rotary encoder, and employing three linear array cameras for coordinated image acquisition. The central camera directly measures the full width of the glass to eliminate Abbe error, and the encoder hardware trigger ensures strict synchronization between image acquisition and glass movement, avoiding image distortion. This invention effectively solves the problems of low measurement accuracy, asynchronous triggering, and poor system robustness in existing technologies, achieving high-precision and high-reliability online inspection of the geometric dimensions and surface defects of flat glass.

Description

Technical Field

[0001] This invention relates to the field of image processing, and in particular to an online inspection method and apparatus for flat glass. Background Technology

[0002] Flat glass is a key basic material in industries such as construction, automotive, and electronics. Its dimensional accuracy (length, width, and diagonal differences) and surface defects (ink spots, white pen lines, and missing corners) directly affect product quality. Currently, online inspection of flat glass production lines mainly faces the following technical challenges:

[0003] 1. Limited Measurement Accuracy: Traditional dual-camera systems calculate glass width indirectly, and their measurement accuracy is limited by the calibration accuracy of the distance between the two cameras. Any mechanical installation error or thermal deformation will introduce Abbe error.

[0004] 2. Poor data acquisition synchronization: Existing systems mostly use software timed triggering or simple photoelectric switch triggering. When the conveyor belt speed fluctuates or the glass slips, the acquired images will be stretched or compressed, which will seriously affect the measurement accuracy and the reliability of defect identification. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a high-precision online inspection method and apparatus for flat glass that can eliminate Abbe error and achieve hardware-level synchronous triggering.

[0006] To achieve the above objectives, the present invention provides an online inspection method for flat glass, comprising:

[0007] Step 1: Use an image acquisition module to acquire an image of the flat glass to be inspected. The image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction. The field of view of the second line array camera completely covers the width direction of the flat glass to be inspected.

[0008] Step 2: Stitch together the images captured by the first line scan camera, the second line scan camera, and the third line scan camera to obtain a complete image;

[0009] Step 3: Preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera;

[0010] Step 4: Calculate the actual width of the flat glass to be tested using formula (1):

[0011] (1);

[0012] Where W is the actual width. and The edge coordinates in the image acquired by the second linear scan camera. This is the pixel equivalent coefficient for the second linear array camera;

[0013] Step 5: Calculate the actual length of the flat glass to be tested using formula (2):

[0014] (2);

[0015] Where L is the actual length. and Δ These represent the pixel span of the glass along its length in the images captured by the first and third linear scan cameras, respectively. and This represents the pixel equivalent coefficient for the corresponding line scan camera.

[0016] Optionally, step 1 includes:

[0017] Step 11: When the photoelectric sensor detects the leading edge of the flat glass to be detected, it generates a trigger signal and sends it to the controller;

[0018] Step 12: After receiving the trigger signal, the controller immediately latches the current rotary encoder count value C1;

[0019] Step 13: Calculate the target trigger position using formula (3) based on the preset calibration distance S:

[0020] (3);

[0021] Where P is the encoder resolution, S is the distance from the photoelectric sensor to the camera, and Δ is the trigger frequency division coefficient;

[0022] Step 14: When the count value reaches or exceeds the target trigger position. At that time, a start acquisition command was sent to the three cameras;

[0023] Step 15: Each camera synchronously acquires image data of the flat glass to be inspected under the pulse trigger of the encoder.

[0024] Optionally, step 2 includes: Step 21: During image acquisition, at each acquisition moment triggered by the encoder pulse, the system records the precise count value of the encoder at that moment and uses this value as the common position label of the three lines of image data acquired simultaneously by the three cameras at that moment.

[0025] Step 22: In the stitching stage, based on the encoder count value as the sequence reference, the three rows of image data acquired under the same encoder count value are arranged in the spatial order corresponding to the first, second, and third line scan cameras to form a complete horizontal strip image.

[0026] Step 23: Arrange all the horizontal strip images in ascending order of encoder count values ​​to reconstruct a complete two-dimensional digital image that corresponds one-to-one with the physical region of the glass.

[0027] Step 24: Obtain the pixel-level mapping relationship between the images of each camera through pre-calibration, and fine-tune the image rows during stitching to ensure smooth and accurate stitching in the overlapping areas of the camera's field of view.

[0028] Optionally, step 3 includes: establishing a unified world coordinate system using coordinate transformation formula (4):

[0029] (4);

[0030] in, , , These represent the coordinates of points in the world coordinate system; T represents... , , These represent the coordinates in the camera coordinate system; r represents the rotation parameter; the subscript numbers represent the linear camera number; R represents the rotation matrix; and t is the translation vector.

[0031] The perspective projection from the camera coordinate system to the image coordinate system is performed according to formula (5):

[0032] (5);

[0033] in, , These represent the coordinates in the image coordinate system; , These represent the focal lengths of the camera in the x and y directions, respectively. These represent the coordinates of the camera principal point (i.e., the intersection of the optical axis and the image plane) in the image pixel coordinate system; the complete camera projection model is shown in formula (6):

[0034] (6);

[0035] in, Let S be the camera intrinsic parameter matrix, and S be the scale factor.

[0036] Optionally, the online detection method for flat glass further includes: Step 6: Calculate the diagonal length of the flat glass to be detected according to formula (7): The specific process is as follows: Perform image analysis on the complete two-dimensional image obtained in step 2, and determine the pixel coordinates of the four corner points of the actual physical contour of the glass plate to be detected in the image coordinate system through edge detection and corner point recognition algorithms. Select one pair of corner points, for example, the physical upper right corner point ( , ) and the physical bottom left corner point ( , Substituting into formula (7) to calculate the diagonal length can be used to calculate the diagonal difference, which is a key quality indicator for evaluating the rectangularity and shape accuracy of the glass plate. Its value cannot be accurately obtained by measuring the length and width alone.

[0037] (7);

[0038] in, , The coordinates of the top right corner point, , The coordinates of the bottom left corner are given. Combined with the system's pixel equivalent (obtainable through calibration), the pixel length D can be converted into the actual physical diagonal length. Similarly, the length of the other diagonal can be calculated; the difference between the two is the key indicator for evaluating the accuracy of the glass shape—the diagonal difference.

[0039] On the other hand, the present invention also provides an online inspection device for flat glass, comprising:

[0040] An image acquisition module is used to acquire images of the flat glass to be inspected. The image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction, wherein the field of view of the second line array camera completely covers the width direction of the flat glass to be inspected.

[0041] The image stitching module is used to stitch together the images acquired by the first line array camera, the second line array camera and the third line array camera to obtain a complete image;

[0042] The preprocessing module is used to preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera;

[0043] The first calculation module is used to calculate the actual width of the flat glass to be tested using formula (1):

[0044] (1);

[0045] Where W is the actual width. and The edge coordinates in the image acquired by the second linear scan camera. This is the pixel equivalent coefficient for the second linear array camera;

[0046] The second calculation module is used to calculate the actual length of the flat glass to be tested using formula (2):

[0047] (2);

[0048] Where L is the actual length. and Δ These represent the pixel span of the glass along its length in the images captured by the first and third linear scan cameras, respectively. and This represents the pixel equivalent coefficient for the corresponding line scan camera.

[0049] Optionally, the online inspection device for flat glass further includes: a third calculation unit, used to calculate the diagonal length of the flat glass to be inspected according to formula (7): The specific process is as follows: image analysis is performed on the complete two-dimensional image obtained in step 2, and the pixel coordinates of the four corner points of the actual physical contour of the glass plate to be inspected in the image coordinate system are determined by edge detection and corner point recognition algorithms. Select one pair of corner points, for example, the physical upper right corner point ( , ) and the physical bottom left corner point ( , Substituting into formula (7) to calculate the diagonal length can be used to calculate the diagonal difference, which is a key quality indicator for evaluating the rectangularity and shape accuracy of the glass plate. Its value cannot be accurately obtained by measuring the length and width alone.

[0050] (7);

[0051] in, , The coordinates of the top right corner point, , The coordinates of the bottom left corner are given. Combined with the system's pixel equivalent (obtainable through calibration), the pixel length D can be converted into the actual physical diagonal length. Similarly, the length of the other diagonal can be calculated; the difference between the two is the key indicator for evaluating the accuracy of the glass shape—the diagonal difference.

[0052] The advantages of this invention over existing technologies are as follows: by directly measuring the full width of the glass using a second linear array camera, the Abbe error of traditional dual-camera systems is fundamentally eliminated; in addition, the use of encoder hardware triggering ensures that the acquisition of each line of images is strictly synchronized with the actual position of the glass, completely avoiding image stretching or compression caused by conveyor belt speed fluctuations; furthermore, the three-camera layout forms a natural redundancy backup, and when any camera fails, the system can automatically switch to a degraded operation mode, still able to complete the detection of key parameters, significantly improving the availability of the system. Attached Figure Description

[0053] Figure 1 This is a flowchart of an online inspection method for flat glass provided by the present invention;

[0054] Figure 2 This is a schematic diagram of the layout of the image acquisition module provided by the present invention;

[0055] Figure 3 This is a timing diagram of the photoelectric sensor and encoder provided by the present invention;

[0056] Figure 4 This is a schematic diagram of coordinate system transformation provided by the present invention;

[0057] Figure 5 This is a structural diagram of an online inspection device for flat glass. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0059] Reference Figure 1 This embodiment provides an online inspection method for flat glass, including the following steps:

[0060] Step 1: Use the image acquisition module to acquire an image of the flat glass to be inspected.

[0061] like Figure 2 As shown, in this embodiment, the image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction. The field of view of the second line array camera completely covers the width direction of the flat glass to be detected. In this way, the second line array camera directly measures the full width of the glass, fundamentally eliminating the Abbe error of the traditional dual-camera system.

[0062] The specific image acquisition process is as follows:

[0063] Step 11: When the photoelectric sensor detects the leading edge of the flat glass to be detected, it generates a trigger signal and sends it to the controller; for example... Figure 3 As shown,

[0064] Step 12: After receiving the trigger signal, the controller immediately latches the current rotary encoder count value C1.

[0065] Step 13: Calculate the target trigger position using formula (3) based on the preset calibration distance S:

[0066] (3);

[0067] Where P is the encoder resolution, S is the distance from the photoelectric sensor to the camera, and Δ is the trigger frequency division coefficient.

[0068] Step 14: When the count value reaches or exceeds the target trigger position. At that time, a start data acquisition command was sent to the three cameras.

[0069] Step 15: Each camera synchronously acquires image data of the flat glass to be inspected under the pulse trigger of the encoder.

[0070] This detection method uses encoder hardware triggering to ensure that the acquisition of each line of images is strictly synchronized with the actual position of the glass, completely avoiding image stretching or compression caused by fluctuations in conveyor belt speed.

[0071] Step 2: Stitch together the images captured by the first line array camera, the second line array camera, and the third line array camera to obtain a complete image.

[0072] Specifically, such as Figure 2 As shown, it includes the following steps:

[0073] Step 21: Location Synchronization Binding: During image acquisition, at each acquisition moment triggered by the encoder pulse, the system records the precise count value of the encoder at that moment and uses this value as the common location label for the three lines of image data acquired simultaneously by the three cameras at that moment.

[0074] Step 22: Row Alignment and Combination: In the stitching stage, the system uses the encoder count value as the sequence reference and arranges the three rows of image data acquired under the same encoder count value according to the spatial order (along the width direction) corresponding to the first, second and third cameras to combine them into a complete horizontal strip image.

[0075] Step 23: Two-dimensional image reconstruction: Arrange all horizontal strip images sequentially according to the encoder count values ​​from smallest to largest, thus reconstructing a complete two-dimensional digital image that corresponds one-to-one with the physical area of ​​the glass. Since the acquisition of all rows is triggered by the same encoder hardware, the image is distorted in the direction of motion.

[0076] Step 24: Spatial registration: Through pre-calibrated cameras, obtain the pixel-level mapping relationship between images from each camera. During stitching, fine-tune the image rows to ensure smooth and accurate stitching in the overlapping areas of the camera's field of view.

[0077] Step 3: Preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera.

[0078] Specifically, such as Figure 4 As shown, a unified world coordinate system is established using coordinate transformation formula (4):

[0079] (4);

[0080] in, , , These represent the coordinates of points in the world coordinate system; T represents... , , These represent the coordinates in the camera coordinate system; r represents the rotation parameter; the subscript numbers represent the linear camera number; R represents the rotation matrix; and t is the translation vector.

[0081] The perspective projection from the camera coordinate system to the image coordinate system is performed according to formula (5):

[0082] (5);

[0083] in, , These represent the coordinates in the image coordinate system; , These represent the focal lengths of the camera in the x and y directions, respectively. These represent the coordinates of the camera principal point (i.e., the intersection of the optical axis and the image plane) in the image pixel coordinate system; the complete camera projection model is shown in formula (6):

[0084] (6);

[0085] in, Let S be the camera intrinsic parameter matrix, and S be the scale factor.

[0086] Step 4: Calculate the actual width of the flat glass to be tested using formula (1):

[0087] (1);

[0088] Where W is the actual width. and The edge coordinates in the image acquired by the second linear scan camera. This represents the pixel equivalent coefficient of the second linear array camera.

[0089] Step 5: Calculate the actual length of the flat glass to be tested using formula (2):

[0090] (2);

[0091] Where L is the actual length. and Δ These represent the pixel span of the glass along its length in the images captured by the first and third linear scan cameras, respectively. and This represents the pixel equivalent coefficient for the corresponding line scan camera.

[0092] This method effectively solves the problems of low measurement accuracy, asynchronous triggering, and poor system robustness in existing technologies, and realizes high-precision and high-reliability online detection of geometric dimensions and surface defects of flat glass.

[0093] In some other embodiments, the online detection method for flat glass further includes step 6: calculating the diagonal length of the flat glass to be detected according to formula (7): the specific process is as follows: performing image analysis on the complete two-dimensional image obtained in step 2, and determining the pixel coordinates of the four corner points of the actual physical contour of the glass plate to be detected in the image coordinate system through edge detection and corner point recognition algorithms. Selecting one pair of corner points, for example, the physical upper right corner point ( , ) and the physical bottom left corner point ( , Substituting into formula (7) to calculate the diagonal length can be used to calculate the diagonal difference, which is a key quality indicator for evaluating the rectangularity and shape accuracy of the glass plate. Its value cannot be accurately obtained by measuring the length and width alone.

[0094] (7);

[0095] in, , The coordinates of the top right corner point, , The coordinates of the bottom left corner are given. Combined with the system's pixel equivalent (obtainable through calibration), the pixel length D can be converted into the actual physical diagonal length. Similarly, the length of the other diagonal can be calculated; the difference between the two is the key indicator for evaluating the accuracy of the glass shape—the diagonal difference.

[0096] It should be noted that edge detection and corner recognition algorithms are existing technologies in this field, and therefore will not be elaborated upon here.

[0097] Reference Figure 5 The present invention also provides an online inspection device for flat glass, comprising:

[0098] The image acquisition module 100 is used to acquire images of the flat glass to be inspected. The image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction. The field of view of the second line array camera completely covers the width direction of the flat glass to be inspected. It should be noted that since the specific image acquisition method and process have been described in detail in step 1 of the above-mentioned online detection method for flat glass, they will not be repeated here.

[0099] The image stitching module 200 is used to stitch together the images acquired by the first line array camera, the second line array camera, and the third line array camera to obtain a complete image. It should be noted that since the specific image stitching method and process have been described in detail in step 2 of the above-mentioned online detection method for flat glass, they will not be repeated here.

[0100] The preprocessing module 300 is used to preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera. It should be noted that since the specific preprocessing method and process have been described in detail in step 3 of the above-mentioned online detection method for planar glass, they will not be repeated here.

[0101] The first calculation module 400 is used to calculate the actual width of the flat glass to be tested using formula (1):

[0102] (1);

[0103] Where W is the actual width. and The edge coordinates in the image acquired by the second linear scan camera. , where is the pixel equivalent coefficient of the second linear array camera; it should be noted that since the specific calculation method and process have been explained in detail in step 4 of the above-mentioned online detection method for planar glass, they will not be repeated here.

[0104] The second calculation module 500 is used to calculate the actual length of the flat glass to be tested using formula (2):

[0105] (2);

[0106] Where L is the actual length. and Δ These represent the pixel span of the glass along its length in the images captured by the first and third linear scan cameras, respectively. and This is the pixel equivalent coefficient of the corresponding line scan camera. It should be noted that since the specific calculation method and process have been explained in detail in step 5 of the above-mentioned online detection method for flat glass, they will not be repeated here.

[0107] In some other embodiments, the online inspection device for flat glass further includes: a third calculation unit, used to calculate the diagonal length of the flat glass to be inspected according to formula (7): the specific process is as follows: image analysis is performed on the complete two-dimensional image obtained in step 2, and the pixel coordinates of the four corner points of the actual physical contour of the glass plate to be inspected in the image coordinate system are determined by edge detection and corner point recognition algorithms. One pair of corner points is selected, for example, the physical upper right corner point ( , ) and the physical bottom left corner point ( , Substituting into formula (7) to calculate the diagonal length can be used to calculate the diagonal difference, which is a key quality indicator for evaluating the rectangularity and shape accuracy of the glass plate. Its value cannot be accurately obtained by measuring the length and width alone.

[0108] (7);

[0109] in, , The coordinates of the top right corner point, , The coordinates are for the lower left corner. It should be noted that the specific calculation method and process have been explained in detail in step 7 of the above-mentioned online detection method for planar glass, so they will not be repeated here.

[0110] In some other embodiments, the online inspection device for flat glass further includes a defect identification module for identifying defect features on the surface of the flat glass to be inspected, including ink spots, white pen lines, and missing corners. It should be noted that since the specific defect identification method and process have been described in detail in step 8 of the above-mentioned online inspection method for flat glass, they will not be repeated here.

[0111] In addition, embodiments of the present invention also provide a computer-readable storage medium, wherein the computer-readable storage medium may store a program that, when executed, includes some or all of the steps of any of the online detection methods for flat glass described in the above method embodiments.

[0112] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0113] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0114] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0115] An exemplary flowchart for implementing an online inspection method for flat glass according to an embodiment of the present invention has been described above with reference to the accompanying drawings. It should be noted that the numerous details included in the above description are merely illustrative of the invention and not intended to limit it. In other embodiments of the invention, the method may have more, fewer, or different steps, and the order, inclusion, function, and other relationships between the steps may differ from those described and illustrated.

Claims

1. A method for on-line inspection of flat glass, characterized in that, include: Step 1: Use an image acquisition module to acquire an image of the flat glass to be inspected. The image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction. The field of view of the second line array camera completely covers the width direction of the flat glass to be inspected. Step 2: Stitch together the images captured by the first line scan camera, the second line scan camera, and the third line scan camera to obtain a complete image; Step 3: Preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera; Step 4: Calculate the actual width of the flat glass to be tested using formula (1): （1）； where W is the actual width, and is the edge coordinate in the image captured by the second linear array camera, is the pixel equivalent coefficient of the second linear array camera; Step 5: Calculate the actual length of the flat glass to be tested using formula (2): （2）； wherein L is the actual length, and Δ are the pixel spans of the glass in the length direction in the images captured by the first and third line array cameras, respectively, and are the pixel equivalent coefficients of the respective line array cameras.

2. The online inspection method for flat glass according to claim 1, characterized in that, Step 1 includes: Step 11: When the photoelectric sensor detects the leading edge of the flat glass to be detected, it generates a trigger signal and sends it to the controller; Step 12: After receiving the trigger signal, the controller immediately latches the current rotary encoder count value C1; Step 13: Calculate the target trigger position using formula (3) based on the preset calibration distance S: （3）； Where P is the encoder resolution, S is the distance from the photoelectric sensor to the first linear array camera, and Δ is the trigger frequency division coefficient; Step 14: When the count value reaches or exceeds the target trigger position. At that time, a start acquisition command was sent to the three cameras; Step 15: Each camera synchronously acquires image data of the flat glass to be inspected under the pulse trigger of the encoder.

3. The online inspection method for flat glass according to claim 1, characterized in that, Step 2 includes: Step 21: During image acquisition, at each acquisition moment triggered by the encoder pulse, the system records the precise count value of the encoder at that moment and uses this value as the common position label of the three lines of image data acquired simultaneously by the three cameras at that moment; Step 22: In the stitching stage, based on the encoder count value as the sequence reference, the three rows of image data acquired under the same encoder count value are arranged in the spatial order corresponding to the first, second, and third line scan cameras to form a complete horizontal strip image. Step 23: Arrange all the horizontal strip images in ascending order of encoder count values ​​to reconstruct a complete two-dimensional digital image that corresponds one-to-one with the physical region of the glass. Step 24: Obtain the pixel-level mapping relationship between the images of each camera through pre-calibration, and fine-tune the image rows during stitching to ensure smooth and accurate stitching in the overlapping areas of the camera's field of view.

4. The online inspection method for flat glass according to claim 1, characterized in that, Step 3 includes: A unified world coordinate system is established using coordinate transformation formula (4): （4）； in, , , These represent the coordinates of points in the world coordinate system; T represents... , , These represent the coordinates of the points in the camera coordinate system; r represents the rotation parameter; the subscript numbers represent the linear camera number; R represents the rotation matrix; and t is the translation vector. The perspective projection from the camera coordinate system to the image coordinate system is performed according to formula (5): （5）； in, , These represent the coordinates in the image coordinate system; , These represent the camera's focal length in the x and y directions, respectively; These represent the coordinates of the camera principal point in the image pixel coordinate system; the complete camera projection model is shown in formula (6): （6）； in, Let S be the camera intrinsic parameter matrix, and S be the scale factor.

5. The online inspection method for flat glass according to claim 1, characterized in that, It also includes: Step 6: Perform image analysis on the complete two-dimensional image obtained in Step 2, and determine the pixel coordinates of the four corner points of the actual physical contour of the plane glass to be detected in the image coordinate system through edge detection and corner point recognition algorithms; Select one pair of diagonal points and substitute them into formula (7) to calculate the length of the diagonal; （7）； in, , The coordinates of the top right corner point, , These are the coordinates of the bottom left corner.

6. An online inspection device for flat glass, characterized in that, include: An image acquisition module is used to acquire images of the flat glass to be inspected. The image acquisition module includes a first line array camera, a second line array camera, and a third line array camera arranged sequentially along the glass conveying direction, wherein the field of view of the second line array camera completely covers the width direction of the flat glass to be inspected. The image stitching module is used to stitch together the images acquired by the first line array camera, the second line array camera and the third line array camera to obtain a complete image; The preprocessing module is used to preprocess the complete image to obtain the coordinate parameters acquired by each line scan camera; The first calculation module is used to calculate the actual width of the flat glass to be tested using formula (1): （1）； Where W is the actual width. and The edge coordinates in the image acquired by the second linear scan camera. This represents the pixel equivalent coefficient of the second linear array camera. The second calculation module is used to calculate the actual length of the flat glass to be tested using formula (2): （2）； Where L is the actual length. and Δ These represent the pixel span of the glass along its length in the images captured by the first and third linear scan cameras, respectively. and This represents the pixel equivalent coefficient for the corresponding line scan camera.

7. The online inspection device for flat glass according to claim 6, characterized in that, Also includes: The third calculation unit is used to perform image analysis on the complete two-dimensional image obtained in step 2, and determine the pixel coordinates of the four corner points of the actual physical contour of the plane glass to be detected in the image coordinate system through edge detection and corner point recognition algorithms. Select one pair of diagonal points and substitute them into formula (7) to calculate the length of the diagonal; （7）； in, , The coordinates of the top right corner point, , These are the coordinates of the bottom left corner.

8. A computer-readable storage medium, comprising a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the online inspection method for flat glass as described in any one of claims 1 to 5.

Images