A method and device for defect detection in float glass production
Through the hierarchical detection process and image processing technology, combined with light incident and clamping flip device, the efficient and accurate defect detection of float glass is achieved, and the problems of low detection efficiency and inaccurate results in the existing technology are solved, and the production efficiency is improved .
Patent Information
- Application Number
- CN202411264852.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-10
AI Technical Summary
The existing floating glass defect detection methods are inefficient, and the reflection and refraction of light sources lead to inaccurate detection results, making it difficult to effectively eliminate unqualified products.
The hierarchical detection process is adopted to first detect whether there are scratches on the end surface of the float glass. The peripheral edge integrity detection is performed only when the end surface is qualified. Scratches and edges are identified through light incident along the normal and image processing technology, and clamping is used. Automatic detection with the flip device to achieve rapid classification.
It improves inspection efficiency, reduces resource waste, ensures the accuracy and automation of inspection results, promptly eliminates unqualified products, and improves production efficiency.
Smart Images

Figure CN118777212B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of float glass defect detection, and in particular to a defect detection method and device for float glass production. Background Art
[0002] Conventional defect detection of float glass includes whether there are scratches on the glass surface, whether there are missing corners on the glass edge, whether there are bubbles inside the glass, etc. The traditional detection method is to obtain float glass data information, and then analyze whether various defects exist based on the float glass data information, obtain the corresponding defect detection results, and then determine whether the float glass is qualified based on the various defect detection results. The unqualified float glass data information needs to be analyzed multiple times before it can be obtained, and the detection efficiency is low. When the existing light source is irradiated on the glass surface, it is reflected and refracted to form two contours, which interferes with the detection and leads to inaccurate detection results. Summary of the invention
[0003] The object of the present invention is to provide a method and device for defect detection in float glass production, and the technical problem to be solved is to eliminate unqualified products step by step and improve the detection efficiency.
[0004] The present invention is achieved through the following technical solutions:
[0005] A first aspect provides a method for defect detection in float glass production, comprising the following steps:
[0006] S100, after receiving the inspection task, calling from the database the size information of the float glass, the position information of the first collection device for collecting the end surface of the float glass, and the position information of the second collection device for collecting the peripheral surface of the float glass;
[0007] S200, determining a collection position for collecting the end face of the float glass through the position information of the first collection device; after placing the float glass at the collection position, starting the first light source and the first collection device, so that the light of the first light source is incident on the end face of the float glass along the normal line; after successful start-up, collecting image information of the end face of the float glass through the first collection device to obtain an image of the end face of the float glass;
[0008] S300, detecting whether there are scratches on the end surface of the float glass on the end surface image of the float glass, and obtaining an end surface detection result;
[0009] S310, when the end surface detection result shows that there are scratches, the float glass is moved to an unqualified area;
[0010] S320: When the end surface detection result shows that there is no scratch, perform the following steps:
[0011] S321, determining the clamping position of the clamping device when clamping the float glass according to the size information of the float glass; determining the flipping position of the flipping device when flipping the float glass according to the size information of the float glass and the position information of the second acquisition device;
[0012] S322, moving the float glass to a clamping position, and clamping it by a clamping device; after clamping, moving the float glass to a flipping position; after moving to the position, starting the second light source and the second acquisition device, and the light of the second light source is incident on the peripheral surface of the float glass along the normal; after the start is successful, flipping the float glass by the flipping device, and collecting image information of the peripheral surface of the float glass by the second acquisition device during flipping, to obtain an image of the peripheral surface of the float glass;
[0013] S323, detecting whether the edge of the float glass is complete on the float glass peripheral surface image, and obtaining an edge detection result;
[0014] S324, when the edge detection result is complete, move the float glass to a qualified area;
[0015] S325: When the edge detection result is incomplete, the float glass is moved to an unqualified area.
[0016] A graded inspection process is adopted, and the end face scratch inspection is carried out first. Only when the end face inspection is qualified will it enter the next stage of edge integrity inspection. Unqualified products can be eliminated step by step, avoiding the waste of resources for comprehensive inspection of all glass and improving the inspection efficiency. The light of this light source is incident on the float glass along the normal, which solves the problem of inaccurate inspection results caused by the light reflecting and refracting on the glass surface to form two contours.
[0017] The end faces of float glass are inspected for scratches, and unqualified glass with scratches on the end faces is preliminarily screened and removed, avoiding further testing of such obviously defective products, thereby saving time and resources; the float glass that has passed the end face inspection is inspected for circumferential integrity, and unqualified glass with incomplete circumferential surfaces is further screened and removed; after each inspection stage, the float glass is immediately classified and unqualified products are removed in a timely manner, preventing unqualified products from continuing to circulate on the production line and further improving overall production efficiency.
[0018] Further, the end surface of the float glass is detected on the end surface image to determine whether there are scratches on the end surface of the float glass, and an end surface detection result is obtained. The specific steps include:
[0019] S301, filtering the float glass end surface image;
[0020] S302, dividing the filtered float glass end face image into grids to obtain a plurality of unit images; numbering the unit images on the float glass end face image using a two-dimensional array to obtain a first array number;
[0021] S303, extracting the float glass contour and scratches in the unit image in sequence according to the first array number, the specific steps are as follows:
[0022] S304, traversing the pixel points in the above unit image, and calculating the gradient amplitude and gradient angle of each pixel point;
[0023] S305, selecting a central pixel point from the unit image; comparing along the gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points;
[0024] If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is a local maximum, and the central pixel is regarded as an edge point;
[0025] If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0;
[0026] S306, traversing the unit images in the float glass end face image; taking a 3*3 pixel matrix as a unit, connecting edge points in any 3*3 pixel matrix on the float glass end face image to form edge lines;
[0027] S307, determining whether the number of edge lines is greater than 1;
[0028] If the number of the above edge lines is equal to 1, then the edge line is the float glass outline, and the end face detection result is that there is no scratch;
[0029] If the number of the above edge lines is greater than 1, one edge line is the float glass outline, and the other edge line is a scratch, and the end face detection result is that there is a scratch.
[0030] Based on the principles of image processing and edge detection, scratches are identified by calculating the gradient information of pixel points and connecting edge points, which not only improves the accuracy of detection but also realizes automatic scratch recognition and classification.
[0031] The float glass end face image is filtered to reduce the noise in the image, highlight the edge information, and make the subsequent scratch detection more accurate; the image is divided into smaller processing units to reduce the single processing volume of the processing unit; at the same time, the unit images are numbered using a two-dimensional array to obtain a first array number so as to process the unit images in sequence; the edge points are determined by calculating the gradient information of the pixel points of the unit image, and the associated edge points are connected into continuous lines (the associated edge points are edge points located in a 3*3 pixel matrix) to facilitate subsequent judgment and analysis of the number of edge lines, thereby obtaining the end face detection result.
[0032] Furthermore, the filtered float glass end face image is grid-splitting to obtain a plurality of unit images, and the specific steps include:
[0033] S3021, taking the upper left vertex of the float glass end surface image as the first reference point, the lower left vertex as the second reference point, the lower right vertex as the third reference point, and the upper right vertex as the fourth reference point;
[0034] S3022, taking the line connecting the first reference point and the second reference point as a reference, obtaining 2n (n is a positive integer, n>2) columns of pixel points from the above-mentioned float glass end surface image to obtain a first matrix image; taking the line connecting the second reference point and the third reference point as a reference, obtaining 2n rows of pixel points from the above-mentioned float glass end surface image to obtain a second matrix image; taking the line connecting the third reference point and the fourth reference point as a reference, obtaining 2n columns of pixel points from the above-mentioned float glass end surface image to obtain a third matrix image; taking the line connecting the fourth reference point and the first reference point as a reference, obtaining 2n rows of pixel points from the above-mentioned float glass end surface image to obtain a fourth matrix image;
[0035] S3023, cleaning the first matrix image, the second matrix image, the third matrix image and the fourth matrix image, and dividing the cleaned first matrix image, the second matrix image, the third matrix image and the fourth matrix image into unit images of 2n*2n pixels;
[0036] S3024, obtaining a fifth matrix image after removing the first matrix image, the second matrix image, the third matrix image and the fourth matrix image from the float glass end face image; dividing the fifth matrix image into unit images of n*n pixels.
[0037] Since the edge of the float glass end face image contains not only the glass outline but also a blank area, which may be free of objects or may be the outline of other objects, the unit images of the edge of the float glass end face image are divided into larger units, which does not affect the scratch detection results, reduces the number of processing times, and improves the detection efficiency. The steps of dividing the float glass end face image are as follows:
[0038] First, the four vertices of the above-mentioned float glass end face image are used as reference points; then, taking the line connecting the first reference point and the second reference point (i.e., the left side of the float glass end face image) as a reference, 2n columns of pixels (n is a positive integer, and n>2) are uniformly obtained from the float glass end face image to construct a first matrix image; taking the line connecting the second reference point and the third reference point (i.e., the lower side of the image) as a reference, 2n rows of pixels are uniformly obtained from the image to construct a second matrix image; similarly, taking the line connecting the third reference point and the fourth reference point (i.e., the right side of the image) as a reference, 2n columns of pixels are obtained to construct a third matrix image; taking the line connecting the fourth reference point and the fourth reference point (i.e., the lower side of the image) as a reference, 2n rows of pixels are uniformly obtained from the image to construct a second matrix image. A line connecting a reference point (i.e., the upper side of the image) is used as a reference to obtain 2n rows of pixel points and construct a fourth matrix image; since the constructed first, second, third and fourth matrix images have repeated pixel points, the constructed first, second, third and fourth matrix images are cleaned to remove the repeated pixel points; after the cleaning is completed, the first, second, third and fourth matrix images are evenly divided into unit images arranged by 2n rows and 2n columns of pixel arrays; after removing the above four matrix images, the remaining float glass end face image portion is defined as a fifth matrix image, and the fifth matrix image is divided into unit images arranged by n rows and n columns of pixel arrays.
[0039] Furthermore, the central pixel point is selected from the above unit image, and the specific steps include:
[0040] S3051, after removing the pixel points located at the edge of the above-mentioned unit image, a first central pixel point set is obtained;
[0041] S3052, numbering the central pixel points on the unit image using a two-dimensional array to obtain a second array number;
[0042] S3053, selecting central pixel points in sequence according to the second array number;
[0043] Until all the central pixel points in the first central pixel point set are traversed.
[0044] To reduce the edge effect, the outermost circle of pixels in the unit image is removed to ensure that the pixels in the first central pixel set are more likely to represent the real features inside the image; the central pixels are numbered using a two-dimensional array, and during the edge detection process, a nested loop (two layers of loops, one for row index and one for column index) is used to select each central pixel in the first central pixel set in turn; each loop iteration processes one pixel until all pixels are processed.
[0045] Furthermore, after traversing all the central pixels in the above unit image,
[0046] S3054, stitching the unit images according to the first array number; obtaining stitching pixel points at the stitching location and pixel points located outside the stitching pixel points, and forming a stitching image from the stitching pixel points and the peripheral pixel points;
[0047] S3055, after removing the pixel points located at the edge of the stitched image, a second central pixel point set is obtained;
[0048] S3056, numbering the central pixel points using a two-dimensional array to obtain a third array number;
[0049] S3057, selecting central pixel points in sequence according to the third array number;
[0050] Until all the central pixel points in the second central pixel point set are traversed.
[0051] After processing one of the above-mentioned unit images, the processed unit images need to be reassembled. When processing the unit images, in order to reduce the edge effect, the pixels at the edge of the unit images are removed. After reassembly, the pixels at the joint are the removed pixels. After the unit images are spliced, some pixels at the joint are no longer at the edge of the image. Therefore, the central pixels at this location need to be processed to improve the detection accuracy.
[0052] Furthermore, after calculating the above gradient angles, the above gradient angles are quantized to obtain a vertical gradient angle, a 45-degree gradient angle, a horizontal gradient angle, and a 135-degree gradient angle.
[0053] When calculating image gradients, the gradient angle (also called gradient direction) indicates the direction in which the pixel values in the image change most dramatically. Since directly processing continuous gradient angle values may be complicated, the gradient angle is usually quantized to simplify subsequent processing. For each pixel in the image, it is assigned to a corresponding vertical gradient angle, 45-degree gradient angle, horizontal gradient angle, or 135-degree gradient angle according to the quantization range to which its gradient angle belongs.
[0054] Further, whether the edge of the float glass is complete on the float glass peripheral surface image is detected to obtain an edge detection result, and the specific steps include:
[0055] S3231, performing filtering processing on the circumferential surface image of the float glass;
[0056] S3232, dividing the filtered float glass circumferential surface image into grids to obtain a plurality of grid images; numbering the grid images on the float glass circumferential surface image using a two-dimensional array to obtain a fourth array number;
[0057] S3233, extracting the float glass contour and the corner-cut contour in the grid image in sequence according to the fourth array number, the specific steps are as follows:
[0058] S3234, traversing the pixel points in the above grid image, and calculating the gradient amplitude and gradient angle of each pixel point;
[0059] S3235, selecting a central pixel point from the above grid image; comparing along the above gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points;
[0060] If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is a local maximum, and the central pixel is regarded as an edge point;
[0061] If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0;
[0062] S3236, traversing the grid image in the float glass peripheral surface image; taking a 3*3 pixel matrix as a unit, connecting edge points located in any 3*3 pixel matrix to form an edge line;
[0063] S3237, determining whether the number of edge lines is greater than 1;
[0064] If the number of the above edge lines is equal to 1, the edge line is the float glass outline, and the edge detection result is complete;
[0065] If the number of the above edge lines is greater than 1, one edge line is the float glass contour, and the other edge line is the missing corner contour, and the edge detection result is incomplete.
[0066] In order to remove the noise in the above-mentioned float glass peripheral surface image and improve the image quality, various filtering algorithms, such as Gaussian filtering and median filtering, can be used to smooth the image and reduce the noise; the filtered image is split into a number of grid images for parallel processing or local feature analysis; the grid images are numbered using a two-dimensional array to ensure that each grid image has a unique identifier; the gradient information of each pixel point in the float glass peripheral surface image is calculated, the local maximum point in the gradient direction is retained as the edge point, the associated edge points are connected to form a continuous edge line, and then the integrity of the edge of the float glass is judged according to the number and shape of the edge lines.
[0067] A second aspect provides a defect detection device for float glass production, the defect detection device adopts the above-mentioned defect detection method;
[0068] The defect detection device comprises:
[0069] A base, wherein a main slide, a first branch slide, a second branch slide, a third branch slide and a fourth branch slide are arranged on the base, the first branch slide and the second branch slide are both connected to the main slide, and the third branch slide and the fourth branch slide are both connected to the first branch slide; the float glass is loaded from the loading end of the main slide;
[0070] A material receiving device, the material receiving device includes a qualified area and an unqualified area; the unloading end of the third branch slide is connected to the qualified area, and the unloading ends of the second branch slide and the fourth branch slide are connected to the unqualified area;
[0071] A first moving device, the first moving device is arranged on the base, the first moving device is used to receive the float glass, and move along the main slide, the first branch slide, the second branch slide, the third branch slide and the fourth branch slide;
[0072] A second moving device, the second moving device is arranged beside the first branch slideway; the second moving device is provided with a flipping device, the flipping device is used to flip the float glass; the flipping device is provided with a clamping device, the clamping device is used to clamp the float glass;
[0073] A first light source and a second light source, both of which are arranged on a base, the first light source is used to illuminate the end surface of the float glass, and the second light source is used to illuminate the peripheral surface of the float glass;
[0074] A first acquisition device, the first acquisition device is arranged on the base, and the first acquisition device is used to acquire image information of the end face of the float glass to obtain an image of the end face of the float glass;
[0075] A second acquisition device, the second acquisition device is arranged on the base, and the second acquisition device is used to acquire image information of the circumferential surface of the float glass to obtain an image of the circumferential surface of the float glass;
[0076] A processor, wherein the processor is connected to the first mobile device, the second mobile device, the first light source, the second light source, the first acquisition device, and the second acquisition device; the processor is used to execute the defect detection method provided by the first aspect.
[0077] The specific steps are as follows:
[0078] S100, calling from a database the size information of the float glass, the position information of a first acquisition device for acquiring data on the end surface of the float glass, and the position information of a second acquisition device for acquiring data on the peripheral surface of the float glass;
[0079] S200, determining a collection position for collecting the end face of the float glass by using the position information of the first collection device; after the float glass is moved to the collection position by using the first moving device, starting the first light source and the first collection device, so that the light of the first light source is incident on the end face of the float glass along the normal line; after successful start-up, collecting image information of the end face of the float glass by using the first collection device to obtain an image of the end face of the float glass;
[0080] S300, detecting whether there are scratches on the end surface of the float glass on the end surface image of the float glass, and obtaining an end surface detection result;
[0081] S310, when the end surface detection result shows that there are scratches, the float glass is moved to an unqualified area;
[0082] S320: When the end surface detection result shows that there is no scratch, perform the following steps:
[0083] S321, determining the clamping position of the clamping device when clamping the float glass according to the size information of the float glass; determining the flipping position of the flipping device when flipping the float glass according to the size information of the float glass and the position information of the second acquisition device;
[0084] S322, moving the float glass to a clamping position by the first moving device, and clamping it by the clamping device; after clamping, moving the float glass to a flipping position by the second moving device; after moving to the position, starting the second light source and the second acquisition device, and the light of the second light source is incident on the peripheral surface of the float glass along the normal; after successful startup, flipping the float glass by the flipping device, and collecting image information of the peripheral surface of the float glass by the second acquisition device during flipping, to obtain an image of the peripheral surface of the float glass;
[0085] S323, detecting whether the edge of the float glass is complete on the float glass peripheral surface image, and obtaining an edge detection result;
[0086] S324, when the edge detection result is complete, move the float glass to a qualified area;
[0087] S325: When the edge detection result is incomplete, the float glass is moved to an unqualified area.
[0088] The above-mentioned base is used as a supporting component; the slide system is used to guide the movement path of the float glass during the inspection process. The float glass enters from the loading end of the main slide to start the inspection process; the first branch slide is used to guide the float glass to the flip device; the second branch slide is used to directly send the float glass that fails the end face inspection to the unqualified area; the third branch slide is used to send the float glass that passes the inspection to the qualified area; the fourth branch slide is used to send the float glass that fails the peripheral surface inspection to the unqualified area; the first moving device is used to receive the float glass and move it along the slide system, which is responsible for turning the float glass The glass is transported from one place to another for inspection or classification; the second moving device is used to adjust the clamping distance of the clamping device, move the float glass from the clamping device to the flipping position, and serve as a supporting component of the flipping device; the flipping device is used to flip the float glass so as to collect image information of its peripheral surface, and the flipping device is provided with a clamping device to ensure that the float glass is stably clamped during the flipping and moving process; the first light source is used to irradiate the end face of the float glass so that the first collecting device collects its image information; the second light source is used to irradiate the peripheral surface of the float glass so that the second collecting device collects its image information.
[0089] Furthermore, the second moving device comprises a horizontal moving assembly, and the horizontal moving assembly comprises supporting plates arranged opposite to each other, and the supporting plates are connected to the base; threaded rods and limiting rods are sleeved on the supporting plates arranged opposite to each other, and the threaded rods are connected to the rotating shaft of the first driver;
[0090] The threaded rod is threadedly connected with a slider, the limit rod passes through the slider, the first driver is used to drive the threaded rod to rotate, the threaded rod drives the slider to move, and the clamping distance of the clamping device is adjusted;
[0091] The slider is provided with a lifting assembly, which includes a second driver and a push rod, wherein the push rod is connected to the rotating shaft of the second driver; the second driver is used to adjust the height of the clamping device; and the flipping device is provided at the telescopic end of the push rod.
[0092] Two supporting plates arranged opposite to each other constitute the base of the horizontal moving component and provide stable support; the threaded rod and the limit rod are both sleeved between the two supporting plates, the threaded rod is connected to the rotating shaft of the first driver, when the first driver is started, its rotating shaft drives the threaded rod to rotate, and the limit rod passes through the slider to guide and prevent the slider from rotating, ensuring that the slider can only move in a straight line; the slider is threadedly connected to the threaded rod, when the threaded rod rotates, due to the constraint of the limit rod, the slider will move along the direction of the threaded rod, by controlling the rotation direction and speed of the first driver, the position of the slider can be adjusted, and then the clamping spacing of the clamping device can be adjusted; a lifting component is arranged on the slider to realize the height adjustment of the clamping device and the flipping device, when the second driver is started, its rotating shaft drives the push rod to perform telescopic movement, the telescopic end of the push rod is connected to the flipping device, through the telescopic movement of the push rod, the height of the flipping device and the clamping device can be adjusted to adapt to float glass of different heights or different detection requirements.
[0093] Compared with the prior art, the present invention has the following advantages and beneficial effects:
[0094] A graded inspection process is adopted, and the end face scratch inspection is carried out first. Only when the end face inspection is qualified will it enter the next stage of edge integrity inspection. Unqualified products can be eliminated step by step, avoiding the waste of resources for comprehensive inspection of all glass and improving the inspection efficiency. The light of this light source is incident on the float glass along the normal, which solves the problem of inaccurate inspection results caused by the light reflecting and refracting on the glass surface to form two contours.
[0095] The end faces of float glass are inspected for scratches, and unqualified glass with scratches on the end faces is preliminarily screened and removed, avoiding further testing of such obviously defective products, thereby saving time and resources; the float glass that has passed the end face inspection is inspected for circumferential integrity, and unqualified glass with incomplete circumferential surfaces is further screened and removed; after each inspection stage, the float glass is immediately classified and unqualified products are removed in a timely manner, preventing unqualified products from continuing to circulate on the production line and further improving overall production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0096] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:
[0097] Figure 1 is the main flow chart of the defect detection method;
[0098] Figure 2 is a simplified diagram of a defect detection device;
[0099] Figure 3 It is a schematic diagram of electrical connection of the defect detection device;
[0100] Figure 4 It is a schematic structural diagram of a second mobile device.
[0101] Marks and corresponding parts names in the attached drawings:
[0102] 10. Main slide; 11. First branch slide; 12. Second branch slide; 13. Third branch slide; 14. Fourth branch slide; 20. Material receiving device; 31. First moving device; 32. Second moving device; 33. Horizontal moving assembly; 331. Support plate; 332. Threaded rod; 333. Limit rod; 334. First drive; 335. Slider; 34. Lifting assembly; 341. Second drive; 342. Push rod; 41. First light source; 42. Second light source; 51. First acquisition device; 52. Second acquisition device; 60. Processor; 70. Clamping device; 80. Flipping device. DETAILED DESCRIPTION
[0103] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.
[0104] Embodiment 1:
[0105] Combination Figure 1 This embodiment 1 provides a method for defect detection in float glass production, comprising the following steps:
[0106] S100, after receiving the inspection task, calling from the database the size information of the float glass, the position information of the first acquisition device 51 for acquiring the end surface of the float glass, and the position information of the second acquisition device 52 for acquiring the peripheral surface of the float glass;
[0107] S200, determining the collection position when collecting the end face of the float glass through the position information of the first collection device 51; after the float glass is vertically placed at the collection position, starting the first light source 41 and the first collection device 51, and the light of the first light source 41 is incident on the end face of the float glass along the normal line (to achieve the light incident along the normal line, two convex lenses with equal focal lengths can be used, and the light emitted by the light source passes through the first convex lens and is collected at the focal length of the first convex lens, and the focus of the second convex lens is located at the focus of the first convex lens, and the collected light becomes parallel light after passing through the second convex lens); after successful startup, the image information of the end face of the float glass is collected through the first collection device 51 to obtain the end face image of the float glass;
[0108] S300, detecting whether there are scratches on the end surface of the float glass on the end surface image of the float glass, and obtaining an end surface detection result;
[0109] S310, when the end surface detection result shows that there are scratches, the float glass is moved to an unqualified area;
[0110] S320: When the end surface detection result shows that there is no scratch, perform the following steps:
[0111] S321, determining the clamping position of the clamping device 70 when clamping the float glass according to the size information of the float glass; determining the flipping position of the flipping device 80 when flipping the float glass according to the size information of the float glass and the position information of the second acquisition device 52;
[0112] S322, moving the float glass to a clamping position, and clamping it by a clamping device 70; after clamping, moving the float glass to a flipping position; after moving to the position, starting the second light source 42 and the second acquisition device 52, and the light of the second light source 42 is incident on the peripheral surface of the float glass along the normal; after successful startup, flipping the float glass by the flipping device 80, and collecting image information of the peripheral surface of the float glass by the second acquisition device 52 during flipping, to obtain an image of the peripheral surface of the float glass;
[0113] S323, detecting whether the edge of the float glass is complete on the float glass peripheral surface image, and obtaining an edge detection result;
[0114] S324, when the edge detection result is complete, move the float glass to a qualified area;
[0115] S325: When the edge detection result is incomplete, the float glass is moved to an unqualified area.
[0116] A graded inspection process is adopted, and the end face scratch inspection is carried out first. Only when the end face inspection is qualified will it enter the next stage of edge integrity inspection. Unqualified products can be eliminated step by step, avoiding the waste of resources for comprehensive inspection of all glass and improving the inspection efficiency. The light of this light source is incident on the float glass along the normal, which solves the problem of inaccurate inspection results caused by the light reflecting and refracting on the glass surface to form two contours.
[0117] The end faces of float glass are inspected for scratches, and unqualified glass with scratches on the end faces is preliminarily screened and removed, avoiding further testing of such obviously defective products, thereby saving time and resources; the float glass that has passed the end face inspection is inspected for circumferential integrity, and unqualified glass with incomplete circumferential surfaces is further screened and removed; after each inspection stage, the float glass is immediately classified and unqualified products are removed in a timely manner, preventing unqualified products from continuing to circulate on the production line and further improving overall production efficiency.
[0118] An application scenario for reference: A factory produces a batch of float glass, and detects qualified products in the batch of float glass at the factory stage, and loads the qualified products for shipment; the size of the batch of float glass is 2000*1500*3mm, the position of the first collection device 51 is fixed, and the distance from the first collection device 51 is When the first acquisition device 51 has an acquisition range of ,use Determine the distance between the float glass and the first collection device 51 when collecting the end face of the float glass , so as to determine the collection location;
[0119] After the float glass is moved to the collection position, the first light source 41 and the first collection device 51 are started, and the image information of the end face of the float glass is collected by the first collection device 51 at the collection position to obtain the end face image of the float glass; the end face image of the float glass is detected, and if the end face detection is unqualified, the float glass is moved to the unqualified area; if the end face detection is qualified, the circumferential surface integrity detection is performed, and the specific steps are as follows: according to the size information of the float glass (2000*1500*3mm), the height of the clamping center is determined mm and minimum height of the flip center mm; the position of the second acquisition device 52 is fixed and the height of the second acquisition device 52 is ; The distance from the second acquisition device 52 is When the collection range of the second collection device 52 is ,use Determine the distance between the float glass and the second collection device 52 when collecting the periphery of the float glass , in order to determine the maximum height of the flip center ; The above heights are all measured based on the slide.
[0120] The float glass is moved to a clamping position and clamped by a clamping device 70. The float glass is then moved to a flipping position. The second light source 42 and the second acquisition device 52 are started, and the light from the second light source 42 is incident on the circumference of the float glass along the normal. The float glass is flipped by a flipping device 80, and the image information of the circumference of the float glass is collected by the second acquisition device 52 during flipping to obtain an image of the circumference of the float glass. The image of the circumference of the float glass is detected. If the circumference detection is unqualified, the float glass is moved to an unqualified area. If the detection is qualified, the float glass is moved to a qualified area.
[0121] Embodiment 2:
[0122] On the basis of Example 1, whether there is a scratch on the end face of the float glass in the end face image of the float glass is detected to obtain the end face detection result, and the specific steps include:
[0123] S301, filtering the float glass end surface image;
[0124] S302, dividing the filtered float glass end face image into grids to obtain a plurality of unit images; numbering the unit images on the float glass end face image using a two-dimensional array to obtain a first array number;
[0125] S303, extracting the float glass contour and scratches in the unit image in sequence according to the first array number, the specific steps are as follows:
[0126] S304, traversing the pixel points in the above unit image, and calculating the gradient amplitude and gradient angle of each pixel point;
[0127] S305, selecting a central pixel point from the unit image; comparing along the gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points;
[0128] If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is a local maximum, and the central pixel is regarded as an edge point;
[0129] If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0, and the non-edge point is excluded;
[0130] S306, traversing the unit images in the float glass end face image; taking a 3*3 pixel matrix as a unit, connecting edge points in any 3*3 pixel matrix on the float glass end face image to form edge lines;
[0131] S307, determining whether the number of edge lines is greater than 1;
[0132] If the number of the above edge lines is equal to 1, then the edge line is the float glass outline, and the end face detection result is that there is no scratch;
[0133] If the number of the above edge lines is greater than 1, one edge line is the float glass outline, and the other edge line is a scratch, and the end face detection result is that there is a scratch.
[0134] Based on the principles of image processing and edge detection, scratches are identified by calculating the gradient information of pixel points and connecting edge points, which not only improves the accuracy of detection but also realizes automatic scratch recognition and classification.
[0135] The float glass end face image is filtered to reduce the noise in the image, highlight the edge information, and make the subsequent scratch detection more accurate; the image is divided into smaller processing units to reduce the single processing volume of the processing unit; at the same time, the unit images are numbered using a two-dimensional array to obtain a first array number so as to process the unit images in sequence; the edge points are determined by calculating the gradient information of the pixel points of the unit image, and the associated edge points are connected into continuous lines (the associated edge points are edge points located in a 3*3 pixel matrix) to facilitate subsequent judgment and analysis of the number of edge lines, thereby obtaining the end face detection result.
[0136] Embodiment 3:
[0137] Based on Example 2, the image of the end face of the float glass has 2n multiple pixel points arranged in the length direction and 2n multiple pixel points arranged in the width direction;
[0138] The filtered float glass end face image is meshed to obtain a number of unit images. The specific steps include:
[0139] S3021, taking the upper left vertex of the float glass end surface image as the first reference point, the lower left vertex as the second reference point, the lower right vertex as the third reference point, and the upper right vertex as the fourth reference point;
[0140] S3022, taking the line connecting the first reference point and the second reference point as a reference, obtaining 2n (n is a positive integer, n>2) columns of pixel points from the above-mentioned float glass end surface image to obtain a first matrix image; taking the line connecting the second reference point and the third reference point as a reference, obtaining 2n rows of pixel points from the above-mentioned float glass end surface image to obtain a second matrix image; taking the line connecting the third reference point and the fourth reference point as a reference, obtaining 2n columns of pixel points from the above-mentioned float glass end surface image to obtain a third matrix image; taking the line connecting the fourth reference point and the first reference point as a reference, obtaining 2n rows of pixel points from the above-mentioned float glass end surface image to obtain a fourth matrix image;
[0141] S3023, cleaning the first matrix image, the second matrix image, the third matrix image and the fourth matrix image, and dividing the cleaned first matrix image, the second matrix image, the third matrix image and the fourth matrix image into unit images of 2n*2n pixels;
[0142] S3024, obtaining a fifth matrix image after removing the first matrix image, the second matrix image, the third matrix image and the fourth matrix image from the float glass end face image; dividing the fifth matrix image into unit images of n*n pixels.
[0143] Since the edge of the float glass end face image contains not only the glass outline but also a blank area, which may be free of objects or may be the outline of other objects, the unit images of the edge of the float glass end face image are divided into larger units, which does not affect the scratch detection results, reduces the number of processing times, and improves the detection efficiency. The steps of dividing the float glass end face image are as follows:
[0144] First, the four vertices of the above-mentioned float glass end face image are used as reference points; then, taking the line connecting the first reference point and the second reference point (i.e., the left side of the float glass end face image) as a reference, 2n columns of pixels (n is a positive integer, and n>2) are uniformly obtained from the float glass end face image to construct a first matrix image; taking the line connecting the second reference point and the third reference point (i.e., the lower side of the image) as a reference, 2n rows of pixels are uniformly obtained from the image to construct a second matrix image; similarly, taking the line connecting the third reference point and the fourth reference point (i.e., the right side of the image) as a reference, 2n columns of pixels are obtained to construct a third matrix image; taking the line connecting the fourth reference point and the fourth reference point (i.e., the lower side of the image) as a reference, 2n rows of pixels are uniformly obtained from the image to construct a second matrix image. A line connecting a reference point (i.e., the upper side of the image) is used as a reference to obtain 2n rows of pixel points and construct a fourth matrix image; since the constructed first, second, third and fourth matrix images have repeated pixel points, the constructed first, second, third and fourth matrix images are cleaned to remove the repeated pixel points; after the cleaning is completed, the first, second, third and fourth matrix images are evenly divided into unit images arranged by 2n rows and 2n columns of pixel arrays; after removing the above four matrix images, the remaining float glass end face image portion is defined as a fifth matrix image, and the fifth matrix image is divided into unit images arranged by n rows and n columns of pixel arrays.
[0145] Embodiment 4:
[0146] On the basis of Example 2, the central pixel point is selected from the above unit image, and the specific steps include:
[0147] S3051, after removing the pixel points located at the edge of the above-mentioned unit image, a first central pixel point set is obtained;
[0148] S3052, numbering the central pixel points on the unit image using a two-dimensional array to obtain a second array number;
[0149] S3053, selecting central pixel points in sequence according to the second array number;
[0150] Until all the central pixel points in the first central pixel point set are traversed.
[0151] To reduce the edge effect, the outermost circle of pixels in the unit image is removed to ensure that the pixels in the first central pixel set are more likely to represent the real features inside the image; the central pixels are numbered using a two-dimensional array, and during the edge detection process, a nested loop (two layers of loops, one for row index and one for column index) is used to select each central pixel in the first central pixel set in turn; each loop iteration processes one pixel until all pixels are processed.
[0152] An application scenario for reference: Assume that each unit image is m*m pixels. After removing the outermost circle of pixels, the remaining (m-1)*(m-1) pixels are the central pixel set.
[0153] A two-dimensional array (second array) is used to number the pixels in the first central pixel set, and two indexes (row index and column index) are used to uniquely identify each pixel, and the index values constitute the second array number B[i][j]. According to the second array number, a nested loop is used to select each pixel in the first central pixel set in turn, and in each loop, the required calculation or detection operation can be performed on the currently selected pixel.
[0154] In a specific embodiment, after traversing all the central pixels in the above unit image,
[0155] S3054, stitching the unit images according to the first array number; obtaining stitching pixel points at the stitching location and pixel points located outside the stitching pixel points, and forming a stitching image from the stitching pixel points and the peripheral pixel points;
[0156] S3055, after removing the pixel points located at the edge of the stitched image, a second central pixel point set is obtained;
[0157] S3056, numbering the central pixel points using a two-dimensional array to obtain a third array number;
[0158] S3057, selecting central pixel points in sequence according to the third array number;
[0159] Until all the central pixel points in the second central pixel point set are traversed.
[0160] After processing one of the above-mentioned unit images, the processed unit images need to be reassembled. When processing the unit images, in order to reduce the edge effect, the pixels at the edge of the unit images are removed. After reassembly, the pixels at the joint are the removed pixels. After the unit images are spliced, some pixels at the joint are no longer at the edge of the image. Therefore, the central pixels at this location need to be processed to improve the detection accuracy.
[0161] In a specific embodiment, after calculating the above-mentioned gradient angle, the above-mentioned gradient angle is quantized to obtain a vertical gradient angle (when the gradient angle is close to 0 degrees or 180 degrees, the threshold range is: (-22.5, 0] ∪ (0, 22.5) ∪ (157.5, 180] ∪ (180, -157.5]), a 45-degree gradient angle (when the gradient angle is close to 45 degrees, the threshold range is: [22.5, 67.5) ∪ [-157.5, -122.5)), a horizontal gradient angle (when the gradient angle is close to 90 degrees, the threshold range is: [67.5, 112.5] ∪ [112.5, -67.5]) and a 135-degree gradient angle (when the gradient angle is close to 135 degrees, the threshold range is: (112.5, 157.5] ∪ [-67.5, -22.5].
[0162] When calculating image gradients, the gradient angle (also called gradient direction) indicates the direction in which the pixel values in the image change most dramatically. Since directly processing continuous gradient angle values may be complicated, the gradient angle is usually quantized to simplify subsequent processing. For each pixel in the image, it is assigned to a corresponding vertical gradient angle, 45-degree gradient angle, horizontal gradient angle, or 135-degree gradient angle according to the quantization range to which its gradient angle belongs.
[0163] Embodiment 5:
[0164] On the basis of Example 1, the image of the circumferential surface of the float glass is arranged with 2n multiple pixels in the length direction and 2n multiple pixels in the width direction; detecting whether the edge of the float glass is complete on the circumferential surface image of the float glass to obtain the edge detection result, the specific steps include:
[0165] S3231, performing filtering processing on the circumferential surface image of the float glass;
[0166] S3232, dividing the filtered float glass circumferential surface image into grids to obtain a plurality of grid images; numbering the grid images on the float glass circumferential surface image using a two-dimensional array to obtain a fourth array number;
[0167] S3233, extracting the float glass contour and the corner-cut contour in the grid image in sequence according to the fourth array number, the specific steps are as follows:
[0168] S3234, traversing the pixel points in the above grid image, and calculating the gradient amplitude and gradient angle of each pixel point;
[0169] S3235, selecting a central pixel point from the above grid image; comparing along the above gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points;
[0170] If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is a local maximum, and the central pixel is regarded as an edge point;
[0171] If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0;
[0172] S3236, traversing the grid image in the float glass peripheral surface image; taking a 3*3 pixel matrix as a unit, connecting edge points located in any 3*3 pixel matrix to form an edge line;
[0173] S3237, determining whether the number of edge lines is greater than 1;
[0174] If the number of the above edge lines is equal to 1, the edge line is the float glass outline, and the edge detection result is complete;
[0175] If the number of the above edge lines is greater than 1, one edge line is the float glass contour, and the other edge line is the missing corner contour, and the edge detection result is incomplete.
[0176] In order to remove the noise in the above-mentioned float glass peripheral surface image and improve the image quality, various filtering algorithms, such as Gaussian filtering and median filtering, can be used to smooth the image and reduce the noise; the filtered image is split into a number of grid images for parallel processing or local feature analysis; the grid images are numbered using a two-dimensional array to ensure that each grid image has a unique identifier; the gradient information of each pixel point in the float glass peripheral surface image is calculated, the local maximum point in the gradient direction is retained as the edge point, the associated edge points are connected to form a continuous edge line, and then the integrity of the edge of the float glass is judged according to the number and shape of the edge lines.
[0177] An application scenario for reference: split the float glass end face image into grids, and use a two-dimensional array to number the unit images on the float glass end face image to obtain the first array number A[i][j], extract the float glass contour and scratches in the unit image from A[0][0], and then extract the unit image of A[0][1]; take the unit image with the first array number A[0][0] as an example, use Calculate the gradient amplitude of each pixel, where Represents pixel The gradient amplitude of Represents the Soble level operator and pixel point The differential of Represents the Soble vertical operator and pixel point Differentiation of Calculate the gradient angle, where Represents pixel The gradient angle.
[0178] Select the pixel numbered B[0][0] in the second array from the unit image As the central pixel, along this pixel The gradient angle of the central pixel of the 3*3 pixel matrix is compared to see whether the gradient amplitude of the central pixel is smaller than the gradient amplitude of the adjacent pixel; if the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is the local maximum, and the central pixel is taken as the edge point; if the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0; then the pixel numbered B[0][1] in the second array is selected. As the central pixel point, along the above pixel points The gradient angle is used to compare whether the gradient amplitude of the central pixel point of the 3*3 pixel matrix is smaller than the gradient amplitude of the adjacent pixel points; the subsequent operation steps are consistent with the above steps until all the central pixel points in the unit image are traversed;
[0179] After traversing the unit images numbered A[0][0] and A[0][1] in the first array, the two unit images are spliced to obtain a larger image, and the spliced pixel points at the splicing position and the pixel points located outside the above spliced pixel points are obtained, and the spliced pixel points and the peripheral pixel points form a spliced image; select the pixel point numbered C[0][0] in the third array from the spliced image As the central pixel, along this pixel The gradient angle of the central pixel of the 3*3 pixel matrix is compared to see whether the gradient amplitude of the central pixel is smaller than the gradient amplitude of the adjacent pixel; if the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel, the gradient amplitude of the central pixel is the local maximum, and the central pixel is taken as the edge point; if the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0; then the pixel numbered B[0][1] in the second array is selected. As the central pixel point, the subsequent operation steps are consistent with the above steps until all the central pixels in the spliced image are traversed.
[0180] After traversing all unit images, the edge points on the float glass end face image located in any 3*3 pixel dot matrix are connected to form edge lines; if the number of the above edge lines is equal to 1, then the edge line is the outline of the float glass, and the end face detection result is that there is no scratch. The float glass is moved to the clamping device for peripheral surface integrity detection, and the detection method is the same as the end face detection method; if the number of the above edge lines is greater than 1, then one edge line is the float glass outline and the other edge line is a scratch. The end face detection result is that there is a scratch, and the float glass is moved to the unqualified area.
[0181] Embodiment 6:
[0182] Combination Figure 2 and Figure 3 , this embodiment 6 provides a defect detection device for float glass production, the defect detection device adopts the above-mentioned defect detection method;
[0183] The defect detection device comprises:
[0184] A base, on which a main slide 10, a first branch slide 11, a second branch slide 12, a third branch slide 13 and a fourth branch slide 14 are arranged, the first branch slide 11 and the second branch slide 12 are both connected to the main slide 10, the third branch slide 13 and the fourth branch slide 14 are both connected to the first branch slide 11; the float glass is loaded from the loading end of the main slide 10;
[0185] The material receiving device 20 includes a qualified area and an unqualified area; the unloading end of the third branch chute 13 is connected to the qualified area, and the unloading ends of the second branch chute 12 and the fourth branch chute 14 are connected to the unqualified area;
[0186] A first moving device 31, which is disposed on a base and is used to receive the float glass and move along the main slide 10, the first branch slide 11, the second branch slide 12, the third branch slide 13 and the fourth branch slide 14;
[0187] A second moving device 32, the second moving device 32 is arranged beside the first branch slide 11; the second moving device 32 is provided with a flipping device 80, the flipping device 80 is used to flip the float glass; the flipping device 80 is provided with a clamping device 70, the clamping device 70 is used to clamp the float glass;
[0188] A first light source 41 and a second light source 42, both of which are disposed on a base, the first light source 41 being used to illuminate the end face of the float glass, and the second light source 42 being used to illuminate the peripheral face of the float glass;
[0189] A first acquisition device 51, the first acquisition device 51 is arranged on the base, and the first acquisition device 51 is used to acquire image information of the end face of the float glass to obtain an image of the end face of the float glass;
[0190] A second acquisition device 52, the second acquisition device 52 is arranged on the base, and the second acquisition device 52 is used to acquire image information of the circumference of the float glass to obtain an image of the circumference of the float glass;
[0191] A processor 60, the processor 60 is connected to the first mobile device 31, the second mobile device 32, the first light source 41, the second light source 42, the first acquisition device 51 and the second acquisition device 52; the processor 60 is used to execute the above-mentioned defect detection method;
[0192] The specific steps are as follows:
[0193] S100, calling from a database the size information of the float glass, the position information of the first acquisition device 51 for acquiring data on the end surface of the float glass, and the position information of the second acquisition device 52 for acquiring data on the peripheral surface of the float glass;
[0194] S200, determining the collection position when collecting the end face of the float glass through the position information of the first collection device 51; after the float glass is moved to the collection position through the first moving device 31, the first light source 41 and the first collection device 51 are started, and the light of the first light source 41 is incident on the end face of the float glass along the normal line; after the start is successful, the image information of the end face of the float glass is collected through the first collection device 51 to obtain the end face image of the float glass;
[0195] S300, detecting whether there are scratches on the end surface of the float glass on the end surface image of the float glass, and obtaining an end surface detection result;
[0196] S310, when the end surface detection result shows that there are scratches, the float glass is moved to an unqualified area;
[0197] S320: When the end surface detection result shows that there is no scratch, perform the following steps:
[0198] S321, determining the clamping position of the clamping device 70 when clamping the float glass according to the size information of the float glass; determining the flipping position of the flipping device 80 when flipping the float glass according to the size information of the float glass and the position information of the second acquisition device 52;
[0199] S322, move the float glass to the clamping position by the first moving device 31, and clamp it by the clamping device 70; after the clamping is completed, move the float glass to the flipping position by the second moving device 32; after the movement is in place, start the second light source 42 and the second acquisition device 52, and the light of the second light source 42 is incident on the peripheral surface of the float glass along the normal; after the start is successful, flip the float glass by the flipping device 80, and collect image information of the peripheral surface of the float glass by the second acquisition device 52 during flipping, so as to obtain the peripheral surface image of the float glass;
[0200] S323, detecting whether the edge of the float glass is complete on the float glass peripheral surface image, and obtaining an edge detection result;
[0201] S324, when the edge detection result is complete, move the float glass to a qualified area;
[0202] S325: When the edge detection result is incomplete, the float glass is moved to an unqualified area.
[0203] The above-mentioned base is used as a supporting component; the slide system is used to guide the movement path of the float glass during the inspection process. The float glass enters from the loading end of the main slide 10 and starts the inspection process; the first branch slide 11 is used to guide the float glass to the flip device 80; the second branch slide 12 is used to directly send the float glass that fails the end face inspection to the unqualified area; the third branch slide 13 is used to send the float glass that passes the inspection to the qualified area; the fourth branch slide 14 is used to send the float glass that fails the peripheral surface inspection to the unqualified area; the first moving device 31 is used to receive the float glass and move it along the slide system, which is responsible for transferring the float glass from one end face to the other end face. The first light source 41 is used to illuminate the end face of the float glass so that the first acquisition device 51 can acquire its image information; the second light source 42 is used to illuminate the peripheral surface of the float glass so that the second acquisition device 52 can acquire its image information.
[0204] Specific embodiments, combined with Figure 4 The second moving device 32 includes a horizontal moving assembly 33, and the horizontal moving assembly 33 includes a supporting plate 331 arranged opposite to each other, and the supporting plate 331 is connected to the base; a threaded rod 332 and a limiting rod 333 are sleeved on the supporting plate 331 arranged opposite to each other, and the threaded rod 332 is connected to the rotating shaft of the first driver 334;
[0205] The threaded rod 332 is threadedly connected with a slider 335, and the limit rod 333 passes through the slider 335. The first driver 334 is used to drive the threaded rod 332 to rotate, and the threaded rod 332 drives the slider 335 to move, so as to adjust the clamping distance of the clamping device 70.
[0206] A lifting assembly 34 is provided on the slider 335, and the lifting assembly 34 includes a second driver 341 and a push rod 342, and the push rod 342 is connected to the rotating shaft of the second driver 341; the second driver 341 is used to adjust the height of the clamping device 70; the flipping device 80 is provided at the telescopic end of the push rod 342; the flipping device 80 can be controlled to rotate by a stepper motor, and the first driver 334, the second driver 341 and the stepper motor are connected to the processor 60.
[0207] The two supporting plates 331 arranged opposite to each other constitute the base of the horizontal moving component 33, providing stable support; the threaded rod 332 and the limiting rod 333 are both sleeved between the two supporting plates 331, and the threaded rod 332 is connected to the rotating shaft of the first driver 334. When the first driver 334 is started, its rotating shaft drives the threaded rod 332 to rotate, and the limiting rod 333 passes through the slider 335 to guide and prevent the slider 335 from rotating, ensuring that the slider 335 can only move in a straight line; the slider 335 is threadedly connected to the threaded rod 332. When the threaded rod 332 rotates, due to the constraint of the limiting rod 333, the slider 335 It will move along the direction of the threaded rod 332. By controlling the rotation direction and speed of the first driver 334, the position of the slider 335 can be adjusted, and then the clamping spacing of the clamping device 70 can be adjusted; a lifting component 34 is set on the slider 335 to achieve height adjustment of the clamping device 70 and the flipping device 80. When the second driver 341 is started, its rotating shaft drives the push rod 342 to perform telescopic movement, and the telescopic end of the push rod 342 is connected to the flipping device 80. Through the telescopic movement of the push rod 342, the height of the flipping device 80 and the clamping device 70 can be adjusted to adapt to float glass of different heights or different detection requirements.
[0208] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for defect detection in float glass production, characterized in that: The following steps are involved: S100, after receiving the inspection task, calling from the database the size information of the float glass, the position information of the first collection device (51) for collecting the end surface of the float glass, and the position information of the second collection device (52) for collecting the peripheral surface of the float glass; S200, determining a collection position when collecting the end face of the float glass through the position information of the first collection device (51); After placing the float glass at a collection position, starting a first light source (41) and a first collection device (51), so that light from the first light source (41) is incident on the end face of the float glass along a normal line; after successful start-up, image information of the end face of the float glass is collected by the first collection device (51) to obtain an image of the end face of the float glass; S300, detecting whether there are scratches on the end surface of the float glass in the end surface image of the float glass, and obtaining an end surface detection result; S310, when the end surface detection result shows that there are scratches, move the float glass to an unqualified area; S320: When the end surface detection result shows that there is no scratch, perform the following steps: S321, determining the clamping position of the clamping device (70) when clamping the float glass based on the size information of the float glass; determining the flipping position of the flipping device (80) when flipping the float glass based on the size information of the float glass and the position information of the second acquisition device (52); S322, moving the float glass to a clamping position, and clamping it by a clamping device (70); After the clamping is completed, the float glass is moved to a flipping position; After being moved into position, the second light source (42) and the second collection device (52) are started, and the light of the second light source (42) is incident on the peripheral surface of the float glass along the normal line; After the startup is successful, the float glass is flipped by means of a flipping device (80), and during the flipping, image information of the circumference of the float glass is collected by means of a second collecting device (52) to obtain an image of the circumference of the float glass; S323, detecting whether the edge of the float glass on the circumferential surface image of the float glass is complete, and obtaining an edge detection result; S324, when the edge detection result is complete, moving the float glass to a qualified area; S325, when the edge detection result is incomplete, moving the float glass to an unqualified area; Detecting whether there are scratches on the end face of the float glass on the end face image of the float glass to obtain an end face detection result, the specific steps include: S301, performing filtering processing on the float glass end surface image; S302, dividing the filtered float glass end face image into grids to obtain a plurality of unit images; numbering the unit images on the float glass end face image using a two-dimensional array to obtain a first array number; S303, extracting the float glass contour and scratches in the unit image in sequence according to the first array number, the specific steps are as follows: S304, traversing the pixel points in the unit image, and calculating the gradient amplitude and gradient angle of each pixel point; S305, selecting a central pixel point from the unit image; and comparing along the gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points; If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel points, the gradient amplitude of the central pixel point is a local maximum, and the central pixel point is regarded as an edge point; If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0; S306, traversing the unit images in the float glass end face image; taking a 3*3 pixel matrix as a unit, connecting edge points on the float glass end face image located in any 3*3 pixel matrix to form an edge line; S307, determining whether the number of edge lines is greater than 1; If the number of edge lines is equal to 1, the end surface detection result is that there is no scratch; If the number of the edge lines is greater than 1, the end face detection result is that scratches exist.
2. A method for defect detection in float glass production according to claim 1, characterized in that: The filtered float glass end face image is grid-splitting to obtain a plurality of unit images, and the specific steps include: S3021, taking the upper left vertex of the float glass end surface image as the first reference point, the lower left vertex as the second reference point, the lower right vertex as the third reference point, and the upper right vertex as the fourth reference point; S3022, taking the line connecting the first reference point and the second reference point as a reference, obtaining 2n columns of pixel points from the image of the end face of the float glass, wherein n is a positive integer and n>2, to obtain a first matrix image; taking the line connecting the second reference point and the third reference point as a reference, obtaining 2n rows of pixel points from the image of the end face of the float glass, to obtain a second matrix image; taking the line connecting the third reference point and the fourth reference point as a reference, obtaining 2n columns of pixel points from the image of the end face of the float glass, to obtain a third matrix image; taking the line connecting the fourth reference point and the first reference point as a reference, obtaining 2n rows of pixel points from the image of the end face of the float glass, to obtain a fourth matrix image; S3023, cleaning the first matrix image, the second matrix image, the third matrix image and the fourth matrix image, and dividing the cleaned first matrix image, the second matrix image, the third matrix image and the fourth matrix image into unit images of 2n*2n pixels; S3024, obtaining a fifth matrix image after removing the first matrix image, the second matrix image, the third matrix image and the fourth matrix image from the float glass end face image; and dividing the fifth matrix image into unit images of n*n pixels.
3. A method for defect detection in float glass production according to claim 1, characterized in that: Selecting a central pixel point from the unit image includes the following specific steps: S3051, after removing the pixel points located at the edge of the unit image, obtain a first central pixel point set; S3052, numbering the central pixel points on the unit image using a two-dimensional array to obtain a second array number; S3053, selecting central pixel points in sequence according to the second array number; Until all the central pixel points in the first central pixel point set are traversed.
4. A method for defect detection in float glass production according to claim 3, characterized in that: After traversing all the central pixels in the unit image, S3054, stitching the unit images according to the first array number; obtaining stitching pixel points at the stitching location and pixel points located outside the stitching pixel points, and forming a stitching image from the stitching pixel points and the peripheral pixel points; S3055, after removing the pixel points located at the edge of the stitched image, obtain a second central pixel point set; S3056, numbering the central pixel points using a two-dimensional array to obtain a third array number; S3057, selecting central pixel points in sequence according to the third array number; Until all the central pixel points in the second central pixel point set are traversed.
5. A method for defect detection in float glass production according to claim 1, characterized in that: After the gradient angle is calculated, the gradient angle is quantized to obtain a vertical gradient angle, a 45-degree gradient angle, a horizontal gradient angle, and a 135-degree gradient angle.
6. A method for defect detection in float glass production according to claim 1, characterized in that: Detecting whether the edge of the float glass is complete on the float glass peripheral surface image to obtain an edge detection result, the specific steps include: S3231, performing filtering processing on the peripheral surface image of the float glass; S3232, dividing the filtered float glass peripheral surface image into grids to obtain a plurality of grid images; numbering the grid images on the float glass peripheral surface image using a two-dimensional array to obtain a fourth array number; S3233, extracting the float glass contour and the corner-cut contour in the grid image in sequence according to the fourth array number, the specific steps are as follows: S3234, traversing the pixel points in the grid image, and calculating the gradient amplitude and gradient angle of each pixel point; S3235, selecting a central pixel point from the grid image; and comparing along the gradient angle whether the gradient amplitude of the central pixel point is smaller than the gradient amplitude of the adjacent pixel points; If the gradient amplitude of the central pixel is equal to or greater than the gradient amplitude of the adjacent pixel points, the gradient amplitude of the central pixel point is a local maximum, and the central pixel point is regarded as an edge point; If the gradient amplitude of the central pixel is smaller than the gradient amplitude of any adjacent pixel, the gradient amplitude of the central pixel is set to 0; S3236, traversing the grid image in the float glass peripheral surface image; taking a 3*3 pixel matrix as a unit, connecting edge points located in any 3*3 pixel matrix to form an edge line; S3237, determining whether the number of edge lines is greater than 1; If the number of edge lines is equal to 1, the edge detection result is complete; If the number of edge lines is greater than 1, the edge detection result is incomplete.
7. A defect detection device for float glass production, characterized in that: The defect detection device adopts the defect detection method according to any one of claims 1 to 6; The defect detection device comprises: A base, wherein a main slide (10), a first branch slide (11), a second branch slide (12), a third branch slide (13) and a fourth branch slide (14) are arranged on the base, the first branch slide (11) and the second branch slide (12) are both connected to the main slide (10), and the third branch slide (13) and the fourth branch slide (14) are both connected to the first branch slide (11); the float glass is loaded from the loading end of the main slide (10); A material receiving device (20), the material receiving device (20) comprising a qualified area and an unqualified area; the unloading end of the third branch slideway (13) is connected to the qualified area, and the unloading ends of the second branch slideway (12) and the fourth branch slideway (14) are connected to the unqualified area; A first moving device (31), the first moving device (31) is arranged on a base, the first moving device (31) is used to receive float glass and move along a main slide (10), a first branch slide (11), a second branch slide (12), a third branch slide (13) and a fourth branch slide (14); A second moving device (32), the second moving device (32) being arranged beside the first branch slideway (11); a flipping device (80) being arranged on the second moving device (32), the flipping device (80) being used for flipping the float glass; a clamping device (70) being arranged on the flipping device (80), the clamping device (70) being used for clamping the float glass; A first light source (41) and a second light source (42), wherein the first light source (41) and the second light source (42) are both arranged on a base, the first light source (41) is used to illuminate the end surface of the float glass, and the second light source (42) is used to illuminate the peripheral surface of the float glass; A first acquisition device (51), the first acquisition device (51) being arranged on a base, and the first acquisition device (51) being used to acquire image information of an end face of a float glass to obtain an image of the end face of the float glass; A second acquisition device (52), the second acquisition device (52) is arranged on the base, and the second acquisition device (52) is used to acquire image information of the peripheral surface of the float glass to obtain an image of the peripheral surface of the float glass; A processor (60), wherein the processor (60) is connected to a first moving device (31), a second moving device (32), a first light source (41), a second light source (42), a first acquisition device (51) and a second acquisition device (52); the processor (60) is used to execute the defect detection method according to any one of claims 1 to 6.
8. A defect detection device for float glass production according to claim 7, characterized in that: The second moving device (32) comprises a horizontal moving assembly (33), and the horizontal moving assembly (33) comprises supporting plates (331) arranged opposite to each other, wherein the supporting plates (331) are connected to the base; a threaded rod (332) and a limiting rod (333) are sleeved on the supporting plates (331) arranged opposite to each other, and the threaded rod (332) is connected to the rotating shaft of the first driver (334); The threaded rod (332) is threadedly connected with a slider (335), the limiting rod (333) passes through the slider (335), the first driver (334) is used to drive the threaded rod (332) to rotate, and the threaded rod (332) drives the slider (335) to move, so as to adjust the clamping distance of the clamping device (70); The slider (335) is provided with a lifting assembly (34), the lifting assembly (34) comprising a second driver (341) and a push rod (342), the push rod (342) being connected to a rotating shaft of the second driver (341); the second driver (341) is used to adjust the height of the clamping device (70); the flipping device (80) is provided at the telescopic end of the push rod (342).
