A Visual Detection Method, Device and Medium for Beverage Packaging Defects

Abstract

The present invention discloses a visual detection method, device and medium for beverage packaging defects. The method includes determining the rotation angle of a preset reference feature point according to the position of the preset reference feature point in a first target image and the position of the preset reference feature point in a standard target image; and obtaining images of corresponding planes of a packaging container to be tested at a detection imaging position through a rotated camera group to obtain a group of images to be tested; performing defect detection on the group of images to be tested to obtain a defect detection result. When the defect detection result indicates no defect, obtaining the previous defect information or defect information within a preset time, determining a second target image with a suspicious defect area from the group of images to be tested based on the defect type and defect image information, adjusting the size of the detection parameters of the defect detection method, and using the adjusted detection parameters to perform defect detection on the suspicious defect area to obtain a final detection result. It can greatly improve the detection efficiency and detection accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a visual detection method, device and medium for beverage packaging defects. Background Art

[0002] Beverages are liquids for human consumption, including tea, coffee, fruit juice, soda, mineral water, etc. The prepared beverages are filled into containers by a filling machine and then packaged to obtain beverage products. Usually, after packaging is completed, it is necessary to detect the defects of the beverage products and package the beverage products without detected defects for easy transportation and sale.

[0003] However, in the prior art, when detecting the defects of beverage products, images of the beverage products in the conveying state are usually directly acquired and image detection is performed, or after reaching a preset position, the images of the beverage products are directly acquired and image detection is performed based on the images, and when the detection result is that there are no defects, the detection result is directly determined as the final detection result.

[0004] In the above method, due to the different degrees of changes in the position and angle of the beverage products during the conveying process, the position of the image content (such as label content, pattern content, text content, etc.) acquired by the photographing device is quite different from the content position of the preset standard image. In subsequent processing, it often increases the complexity of image processing and the complexity of machine vision algorithms, resulting in an increase in a large amount of computing resources. Moreover, in the above method, when the detection result is that there are no defects, the detection result is directly determined as the final detection result, lacking further detection of the defects that have occurred recently, and it is impossible to accurately know the key cause points corresponding to the defect problems and the accuracy of repair, which is not conducive to the improvement of subsequent processes and cannot achieve the gradual improvement of the finished product quality of the beverage products produced by the production line. Summary of the Invention

[0005] In order to solve at least one of the above-mentioned technical problems, the present invention provides a visual detection method, device and medium for beverage packaging defects.

[0006] In a first aspect, the present invention provides a visual detection method for beverage packaging defects, the method comprising:

[0007] After detecting that the packaging container to be tested moves to the detection imaging position, an image containing the preset reference feature points of the packaging container to be tested is acquired to obtain a first target image;

[0008] According to the positions of the preset reference feature points in the first target image and the positions of the preset reference feature points in the standard target image, the rotation angle of the preset reference feature points is determined;

[0009] Controlling the rotation of the camera group based on the rotation angle of the preset reference feature points, and obtaining images of corresponding planes of the packaging container to be measured at the detection imaging position through the rotated camera group, thereby obtaining a group of images to be measured;

[0010] Performing defect detection based on the group of images to be measured to obtain a defect detection result. When the defect detection result indicates no defect, obtaining the previous defect information or defect information within a preset time. The defect information includes defect type, defect image information, and defect detection method. Based on the defect type and the defect image information, determining a second target image with a suspicious defect area from the group of images to be measured, adjusting the size of the detection parameters of the defect detection method, and using the adjusted detection parameters to perform defect detection on the suspicious defect area to obtain a final detection result; wherein, the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before adjustment.

[0011] Preferably, obtaining an image containing the preset reference feature points of the packaging container to be measured to obtain a first target image of the packaging container to be measured includes:

[0012] Obtaining an image of a corresponding plane of the packaging container to be measured at the detection imaging position through the camera group at the initial position to obtain a first group of images; wherein, the camera group includes four first image acquisition devices arranged in an array at the detection imaging position. When the packaging container to be measured moves to the detection imaging position, the packaging container to be measured is at the focal position of the four first image acquisition devices;

[0013] Determining an image containing the preset reference feature points from the first group of images, and using the image containing the preset reference feature points as the first target image.

[0014] Preferably, determining a first target image containing the preset reference feature points from the first group of images includes:

[0015] Regarding the first image acquisition device corresponding to the standard target image as the first target image acquisition device;

[0016] Detecting the image acquired by the first target image acquisition device to determine whether the preset reference feature points exist;

[0017] If so, determining the image acquired by the first target image acquisition device as the first target image;

[0018] Otherwise, randomly label the remaining first image acquisition devices as the second target image acquisition device, the third target image acquisition device, and the fourth target image acquisition device, and sequentially detect the images acquired by the second target image acquisition device, the third target image acquisition device, and the fourth target image acquisition device to determine whether the preset reference feature points exist until an image with the preset reference feature points is detected, and determine the image with the preset reference feature points as the first target image.

[0019] Preferably, defect detection is performed on the group of images to be measured to obtain a defect detection result, including:

[0020] The camera group includes a first image acquisition device and a second image acquisition device; wherein, the first image acquisition device is used to obtain a front view of the corresponding plane of the packaging container to be measured, and the second image acquisition device is disposed directly above the detection imaging position, and the second image acquisition device is used to obtain a top view of the packaging container to be measured to form a top surface image;

[0021] The group of images to be measured includes a front image group formed by front views of each plane of the packaging container to be measured, and the top surface image;

[0022] Perform local defect detection on the front image group to obtain a local defect detection result; wherein, the local defect detection includes one or more of label defect detection, lid defect detection, and liquid level defect detection;

[0023] When the local defect detection result indicates no defect, compare each image in the front image group and the top surface image with the corresponding preset standard image to determine whether there is a defect, and obtain an overall defect detection result.

[0024] Preferably, performing local defect detection on the front image group to obtain a local defect detection result includes:

[0025] Obtain a region image containing the label region in each image of the front image group to obtain a third image group, generate a label unfolding diagram based on the third image group, and perform defect detection on the label unfolding diagram to obtain a label defect detection result;

[0026] Obtain a region image containing the lid region and the standard liquid level region in each image of the front image group to obtain a fourth image group, and perform lid and liquid level detection on each image in the fourth image group to obtain a lid defect detection result and a liquid level detection result;

[0027] If at least one of the label defect detection result, the cover defect detection result, and the liquid level detection result shows a defect, it is determined that the packaging container to be tested has a local defect; otherwise, it is determined that there is no local defect.

[0028] Preferably, a label unfolded diagram is generated based on the third image group, and defect detection is performed on the label unfolded diagram, including:

[0029] Preprocess each image in the third image group; wherein, the preprocessing includes smoothing filtering and cylindrical label back-projection correction;

[0030] Stitch and fuse the images in the preprocessed third image group to obtain the label unfolded diagram;

[0031] Use the color information of the beverage and the label, and determine whether there is a label defect through the double threshold of the color space and the morphological processing algorithm.

[0032] Preferably, cover and liquid level detection are performed on each image in the fourth image group to obtain the cover defect detection result and the liquid level detection result, including:

[0033] Perform image enhancement processing on each image in the fourth image group;

[0034] Based on the pre-trained positioning model, perform pattern positioning to obtain the pattern area; wherein, the pattern area includes the cover area and the liquid level area;

[0035] Detect the contrast and cover angle of the cover area to obtain the cover defect detection result;

[0036] Detect the liquid level height of the liquid level area to obtain the liquid level detection result.

[0037] Preferably, each image in the front image group and the top image are compared with the corresponding preset standard images to determine whether there are defects, including:

[0038] Preprocess each image in the front image group and the corresponding preset standard images, wherein the preprocessing includes grayscale conversion, denoising, and normalization; use the Canny edge detection algorithm to extract the image edges; calculate the differential image and perform binary processing; count the total number of pixels in the defect area of the binary image; if the total number of pixels in the defect area is greater than the preset defect area threshold, it is determined that the front image group has a defect, otherwise it is determined that the front image group has no defect;

[0039] Convert the top surface image and the corresponding preset standard image into grayscale images; count the number of pixels at each gray level in the grayscale images; calculate the gray mean value and truncate and amplify the gray difference in the defect area; use statistical methods to compare the gray histograms of the two images; compare the feature difference value with a threshold value to determine whether there are defects in the top surface image;

[0040] If at least one of the front surface image group and the top surface image has defects, it is determined that the packaging container to be tested has defects; if not, it is determined that there are no defects.

[0041] In a second aspect, the present invention also provides a visual inspection device for beverage packaging defects, and the device includes:

[0042] An acquisition module, configured to obtain an image including a preset reference feature point of the packaging container to be tested after detecting that the packaging container to be tested moves to the detection imaging position, and obtain a first target image;

[0043] A determination module, configured to determine the rotation angle of the preset reference feature point according to the position of the preset reference feature point in the first target image and the position of the preset reference feature point in the standard target image;

[0044] A control module, configured to control the rotation of the camera group based on the rotation angle of the preset reference feature point, and obtain images of corresponding planes of the packaging container to be tested at the detection imaging position through the rotated camera group, and obtain a group of images to be tested;

[0045] A first detection module, configured to perform defect detection according to the group of images to be tested, and obtain a defect detection result;

[0046] A second detection module, when the defect detection result is that there are no defects, obtains the previous defect information or the defect information within a preset time, the defect information includes the defect type, defect image information, and defect detection method, determines a second target image with a suspicious defect area from the group of images to be tested based on the defect type and the defect image information, adjusts the detection parameters of the defect detection method, and uses the adjusted detection parameters to perform defect detection on the suspicious defect area to obtain a final detection result; wherein, the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before adjustment.

[0047] In a third aspect, the present invention also provides a computer-readable storage medium, in which a computer program is stored, the computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor is caused to execute the visual inspection method for beverage packaging defects as described in any one of the above.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] 1) For the visual inspection method for beverage packaging defects provided by the present invention, after detecting that the packaging container to be inspected moves to the inspection imaging position, an image containing the preset reference feature points of the packaging container to be inspected is acquired to obtain a first target image; according to the positions of the preset reference feature points in the first target image and the positions of the preset reference feature points in the standard target image, the rotation angle of the preset reference feature points is determined; the camera group is controlled to rotate based on the rotation angle of the preset reference feature points, and images of the corresponding planes of the packaging container to be inspected at the inspection imaging position are acquired through the rotated camera group to obtain a group of images to be inspected; according to the present invention, the rotation angle of the preset reference feature points in the first target image corresponding to the standard target image is determined based on the positions of the preset reference feature points in the first target image, and the camera group is controlled to rotate according to this rotation angle, which can reduce the complexity of subsequent image processing and algorithm complexity, thereby improving the processing efficiency of subsequent images and reducing the computing resource cost of image processing.

[0050] 2) Defect detection is performed according to the group of images to be inspected to obtain a defect detection result; since the images in the group of images to be inspected are images of the corresponding planes of the packaging container to be inspected at the inspection imaging position, the present invention detects the images of each plane of the packaging container to be inspected, making the defect detection result more in line with the actual production situation and accurately ensuring the accuracy of the defect detection result, avoiding situations such as missed detection and missing inspection.

[0051] 3) When the defect detection result is that there are no defects, the previous defect information or defect information within a preset time is acquired. The defect information includes defect type, defect image information, and defect detection method. Based on the defect type and defect image information, a second target image with a suspicious defect area is determined from the group of images to be inspected, the detection parameters of the defect detection method are adjusted in size, and the suspicious defect area is defect-detected using the adjusted detection parameters to obtain a final detection result; among them, the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before adjustment. When the defect detection result is that there are no defects, the present invention will acquire the previous defect information or defect information within a preset time, determine a second target image with a suspicious defect area according to the defect information, and adjust the corresponding detection parameters to further improve the detection accuracy, and defect-detect the suspicious defect area using the adjusted detection parameters. Through this detection method, it is possible to accurately determine and precisely detect the defects that have occurred, thereby determining the corresponding repair accuracy, continuously improving the quality of the finished product, and ultimately greatly improving the quality of the finished product in the entire production process.

[0052] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the present disclosure. Brief Description of the Drawings

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background art, the drawings required for use in the embodiments of the present invention or the background art will be described below.

[0054] The drawings herein are incorporated into the specification and form a part of this specification. These drawings show embodiments in accordance with the present disclosure and, together with the specification, are used to illustrate the technical solutions of the present disclosure.

[0055] Figure 1 It is a schematic flowchart of a visual inspection method for beverage packaging defects provided by an embodiment of the present invention;

[0056] Figure 2 It is a schematic working principle diagram of the S20 sub-process of a visual inspection device for beverage packaging defects provided by an embodiment of the present invention;

[0057] Figure 3 It is a schematic diagram of the weight change of the splicing and fusion area in the label detection step of a visual inspection method for beverage packaging defects provided by an embodiment of the present invention;

[0058] Figure 4 It is a schematic diagram of the simple structure of the conveying device and the detection station provided by an embodiment of the present invention;

[0059] Figure 5 It is a schematic diagram of the structure of a visual inspection device for beverage packaging defects provided by an embodiment of the present invention.

[0060] Reference numerals in the drawings: 1, conveying device; 11, guide rod; 2, black light-absorbing cloth; 3, light source device; 4, camera group; 41, first image acquisition device; 42, second image acquisition device; 51, first sensor; 52, second sensor; 6, rotating mechanism; 7, packaging container to be tested. Detailed Description of the Embodiments

[0061] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0062] The terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present invention are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0063] Reference to "embodiment" herein means that a particular feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0064] In addition, for better illustration of the present invention, numerous specific details are given in the following specific implementation manners. Those skilled in the art should understand that the present invention can still be implemented without some specific details. In some instances, methods, means, elements and circuits well known to those skilled in the art are not described in detail so as to highlight the gist of the present invention.

[0065] Please refer to Figure 1 , Figure 1 which is a schematic flow chart of a visual inspection method for beverage packaging defects provided by an embodiment of the present invention. As Figure 1 shown, a visual inspection method for beverage packaging defects includes the following steps:

[0066] S10. After detecting that the packaging container 7 to be tested moves to the detection imaging position, obtain an image containing the preset reference feature points of the packaging container 7 to be tested to obtain a first target image;

[0067] It should be noted that, please refer to Figure 1 , 4 , the packaging container 7 to be tested can be a bottle body. The packaging container 7 to be tested is conveyed to the detection imaging position through a conveying device and conveyed to the next process through the conveying device after detection. A guide rod 11 is arranged in the area of the conveying device close to the detection imaging position so that the packaging container 7 to be tested can be conveyed to the detection imaging position after the position is corrected by the guide rod 11, thereby improving the accuracy of image acquisition.

[0068] In addition, please refer to Figure 4, the detection imaging position includes a detection station, and the periphery and top of the detection station are surrounded by a black light-absorbing cloth 2; a light source device 3 is arranged at the detection station, and the light source device 3 is below the black light-absorbing cloth 2 at the top; the light source device 3 is an annular light source or a dome light source to improve the imaging quality; the camera group 4 is arranged between the black light-absorbing cloths 2 around the periphery and is below the black light-absorbing cloth 2 at the top; the detection station is also sequentially provided with a first sensor 51 and a second sensor 52 at intervals along the conveying direction, and the induction distance between the first sensor 51 and the second sensor 52 is less than or equal to the outer diameter of the measured part of the packaging container 7 to be measured, so as to determine that the packaging container 7 to be measured moves to the detection imaging position when both the first sensor 51 and the second sensor 52 sense the packaging container 7 to be measured, thereby improving the accuracy of image acquisition.

[0069] Preferably, obtaining an image containing the preset reference feature points of the packaging container 7 to be measured to obtain the first target image of the packaging container 7 to be measured includes:

[0070] S11. Obtain an image of the corresponding plane of the packaging container 7 to be measured at the detection imaging position at the initial position through the camera group 4 to obtain a first image group; wherein, the camera group 4 includes four first image acquisition devices 41 arranged in an array at the detection imaging position, and when the packaging container 7 to be measured moves to the detection imaging position, the packaging container 7 to be measured is at the focal position of the four first image acquisition devices 41;

[0071] It should be noted that, please refer to Figure 4 , through the guide rod 11, the first sensor 51 and the second sensor 52, it can be accurately ensured that the packaging container 7 to be measured is at the focal position of the four first image acquisition devices 41, thereby improving the accuracy of image acquisition, and realizing rapid comparison and analysis with the standard target image, improving the detection efficiency, and reducing the processing complexity and algorithm complexity.

[0072] S12. Determine the image containing the preset reference feature points from the first image group, and use the image containing the preset reference feature points as the first target image. Specifically, it includes:

[0073] S121. Use the first image acquisition device 41 corresponding to the standard target image as the first target image acquisition device;

[0074] S122. Detect the image acquired by the first target image acquisition device to determine whether the preset reference feature points exist;

[0075] S123. If so, determine the image acquired by the first target image acquisition device as the first target image;

[0076] S124. If not, randomly mark the remaining first image acquisition devices 41 as the second target image acquisition device, the third target image acquisition device, and the fourth target image acquisition device, and sequentially detect the images acquired by the second target image acquisition device, the third target image acquisition device, and the fourth target image acquisition device to determine whether the preset reference feature points exist until an image with the preset reference feature points is detected, and determine the image with the preset reference feature points as the first target image.

[0077] It should be noted that the preset reference feature points can be set accordingly according to the actual situation of the packaging container 7 to be measured. For example, if the packaging container 7 to be measured is a bottle body with a packaging label attached to the bottle, the unique obvious pattern or mark on the packaging label can be used as the preset reference feature point to quickly and accurately identify its position and obtain the first target image.

[0078] S20. Determine the rotation angle of the preset reference feature point according to the position of the preset reference feature point in the first target image and the position of the preset reference feature point in the standard target image.

[0079] It should be noted that the standard target image is a reference image of the first target image, which contains the preset reference feature points, and the preset reference feature points in the standard target image are located at the center of the standard target image. Since the standard target image is acquired by the first image acquisition device 41 corresponding to the plane, that is, the preset reference feature points in the standard target image are the parts closest to the first image acquisition device 41 corresponding to the plane. That is to say, a straight line is formed by connecting a point on the central axis of the corresponding plane first image acquisition device 41, the preset reference feature points in the standard target image, and the packaging container 7 to be measured. By calculating according to the position of the preset reference feature points in the first target image and the position of the preset reference feature points in the standard target image, the rotation angle of the preset reference feature points in the first target image relative to the preset reference feature points in the standard target image can be accurately calculated. Specifically, as Figure 2 shown, the following steps can be included:

[0080] S21. Construct a plane coordinate axis.

[0081] S22. Draw a circle a on the plane coordinate axis with the radius of the packaging container 7 to be measured corresponding to the preset reference feature points in the standard target image; among them, the center of the circle coincides with the origin of the plane coordinate axis.

[0082] S23, mapping the preset reference feature points in the first target image and the preset reference feature points in the standard target image on the coordinate axis, and calculating the actual length of the first right-angled side b; specifically comprising: calculating the distance between the preset reference feature points in the first target image and the preset reference feature points in the standard target image, and obtaining the mapping length between the two preset reference feature points; marking the actual positions of the two preset reference feature points on the plane coordinate axis according to the preset ratio, and connecting the actual positions of the two preset reference feature points to obtain the first right-angled side, and calculating the actual length of the first right-angled side b based on the plane coordinate axis; wherein the first right-angled side b is tangent to the circle;

[0083] It should be noted that since the position of the preset reference feature point in the first target image is the position of the packaging container 7 to be tested mapped on the plane, and the length calculated based on the image is the length reduced by a preset ratio, it is necessary to calculate according to the preset ratio to obtain the actual length distance formed by the connection of the two preset reference feature points when they are actually mapped on the plane; in addition, this actual length distance is not the actual moving length of the preset reference feature point, so it is necessary to continue to perform the following steps.

[0084] S24, within the circle, with the center of the circle as the starting point, draw a second right-angled side parallel to the first right-angled side based on the actual length of the first right-angled side, the length of the second right-angled side is b ′ Equal to the length of the first right angle side b;

[0085] S25, according to the length b of the second right angle side ′ The radius length c of the packaging container 7 to be tested corresponding to the preset reference feature point is calculated to obtain the length d of the third right-angled side; wherein the second right-angled side b ′ , the third right-angled side d and the radius c of the packaging container 7 to be measured corresponding to the preset reference feature point are connected to form a first triangle;

[0086] S26, calculating the length of the fourth right-angled side e according to the length of the third right-angled side d and the radius length of the packaging container 7 to be measured corresponding to the preset reference feature point; wherein the length of the fourth right-angled side e is equal to the length of the radius c of the packaging container 7 to be measured corresponding to the preset reference feature point minus the length of the third right-angled side d;

[0087] S27, according to the length of the first right-angled side b and the length of the fourth right-angled side e, calculate the actual moving length of the preset reference feature point (that is, the length of the hypotenuse f); wherein, one end of the first right-angled side b and the fourth right-angled side e intersect vertically, and the other ends of the first right-angled side b and the fourth right-angled side e coincide with the two preset reference feature points respectively, and the hypotenuse f is obtained by connecting the other ends of the first right-angled side b and the fourth right-angled side e;

[0088] S28. Calculate the rotation angle of the preset reference feature point based on the actual moving length and the radius length of the packaging container 7 to be measured corresponding to the preset reference feature point. Specifically, substitute the actual moving length and the radius length of the packaging container 7 to be measured corresponding to the preset reference feature point into the cosine theorem to obtain a radian value, and convert the radian into degrees to obtain the rotation angle θ of the preset reference feature point.

[0089] S30. Control the rotation of the camera group 4 based on the rotation angle of the preset reference feature point, and obtain an image of the corresponding plane of the packaging container 7 to be measured at the detection imaging position through the rotated camera group 4 to obtain a group of images to be measured;

[0090] It should be noted that, please refer to Figure 4 , a rotation mechanism 6 is provided at the detection imaging position, the camera group 4 is installed on the rotation mechanism 6, and the rotation mechanism 6 drives the first image acquisition device 41 to rotate according to the radian value or the rotation angle θ of the preset reference feature point, so as to realize the rotation of the camera group 4. The camera group 4 includes a first image acquisition device 41 and a second image acquisition device 42; wherein, the first image acquisition device 41 is used to obtain a front view of the corresponding plane of the packaging container 7 to be measured, the second image acquisition device 42 is arranged directly above the detection imaging position, and the second image acquisition device 42 is used to obtain a top view of the top surface of the packaging container 7 to be measured to form a top surface image, where Figure 4 the dashed circle in

[0091] In this embodiment, the rotation mechanism 6 includes an arc-shaped slide rail, a slider, a micro motor and an arc-shaped rack. A gear meshed with the arc-shaped rack is installed at the driving end of the micro motor. The slider is slidably connected to the arc-shaped slide rail. The micro motor and the first image acquisition device 41 are arranged on the slider; it should be understood that a preset distance is provided between the arc-shaped slide rail and the arc-shaped rack, and the radian of the arc-shaped rack is set equidistantly according to the radian of the arc-shaped slide rail and the preset distance, so as to ensure that the slider can be driven to slide on the arc-shaped slide rail by the operation of the micro motor, and the radian of the arc-shaped slide rail is preset according to needs.

[0092] The group of images to be measured includes a group of front images formed by the front views of the respective planes of the packaging container 7 to be measured, and the top surface image;

[0093] S40. Perform defect detection based on the group of images to be measured to obtain a defect detection result; specifically, it includes:

[0094] S41. Perform local defect detection on the group of front images to obtain a local defect detection result; wherein, the local defect detection includes one or more of label defect detection, lid defect detection and liquid level defect detection; specifically, it includes:

[0095] S411. Obtain the regional images containing the label regions in each image of the front image group to obtain a third image group, generate a label unfolding diagram based on the third image group, and perform defect detection on the label unfolding diagram to obtain a label defect detection result. Specifically, it includes:

[0096] S411-1. Preprocess each image in the third image group. Among them, the preprocessing includes smoothing filtering and cylindrical label back-projection correction.

[0097] It should be noted that the beverage label is a flat label before sleeving. When heat-shrink sleeving, the label is pasted on the beverage bottle, making the label in a cylindrical shape. Therefore, there are certain distortions in the horizontal and vertical directions of the label on the beverage bottle, which causes a certain degree of loss of label image information. Therefore, it is necessary to preprocess each image in the third image group.

[0098] Since the camera will generate perspective distortion according to the distance of the object during imaging, the characteristic of perspective distortion is that the farther the object is from the lens, the smaller the image of the object formed on the camera. Therefore, only the front half of the cylindrical surface can be observed in the image. At this time, through horizontal and vertical correction compensation, the cylindrical label can be approximately unfolded, and then the missing pixel points can be approximately calculated through the bilinear interpolation algorithm.

[0099] S411-2. Stitch and fuse the images in the preprocessed third image group to obtain the label unfolding diagram.

[0100] It should be noted that after correction, image registration is performed on each pair of adjacent labels to find the corresponding position relationship, and then stitching and fusion processing is performed. Among them, the fade-in and fade-out algorithm is a commonly used stitching and fusion method, and its weight changes linearly. The weight is changed to non-linear to make the stitching transition natural, and its linear function weight is improved to a weight based on the sigmoid function. The expression of the sigmoid function is

[0101] Its weight ω s changes as Figure 3 shown. The expression of the weight ω s is Its value range is (0, 1). Among them, i represents the coordinate value of the pixel point in the overlapping area in the horizontal direction of the image, and L and R respectively represent the coordinate values of the left and right boundaries of the overlapping area in the horizontal direction. By adjusting the value of k, the change trend of the sigmoid function can be changed.

[0102] According to the weight ω sThen the two images can be stitched and fused to obtain the unfolded label image, and the calculation formula for the gray value of the pixel points in the overlapping area is as follows: g(x,y) = ω s g A (x,y) + (1 - ω s )g B (x,y); (x,y) ∈ (g A ∩g B ), g A is image A, g A (x,y) is the gray value of the pixel point at position x,y in image A, g B is image B, g B (x,y) is the gray value of the pixel point at position x,y in image B.

[0103] S411-3. Utilize the color information of the beverage and the label, and determine whether there are label defects through the double-threshold of the color space and the morphological processing algorithm.

[0104] It should be noted that when the label of the beverage bottle is shrink-wrapped, the plastic film label may be broken due to the unevenness of the label shrinker, and the area after the break shows the color of the beverage liquid itself. Based on the color information characteristics of the beverage liquid itself and combined with the color of the label, the color of the beverage liquid itself can be extracted, that is, the beverage liquid color is used as the foreground, and the label information is relatively complex and not unified, that is, used as the background. Through two color spaces based on RGB and HSV, the RGB color space is the most commonly used color representation method in computer technology, while the HSV color space is proposed according to the intuitive characteristics of human beings for colors, which is closer to the human perception of colors and has a certain stability against brightness interference in the environment. For the color characteristics of the beverage liquid itself in the color space, the mean values of the three channels in the RGB space of the beverage liquid, namely R m , G m , B m and the mean values of the three channels of HSV, namely H m , S m , V m , are respectively statistically analyzed, and then an upper and lower threshold is designed as:

[0105] (R m ±R) ∩ (G m ±G) ∩ (B m ±B)

[0106] (H m ±R) ∩ (S m ±G) ∩ (V m ±B)

[0107] Then, the selected pixels are connected to form regions, and a closing operation in morphological processing is performed on these regions. Finally, a union operation is performed on these regions to segment the broken label regions.

[0108] S412. Obtain the region images containing the cap region and the standard liquid level region in each image of the front image group to obtain a fourth image group, and perform cap and liquid level detection on each image in the fourth image group to obtain a cap defect detection result and a liquid level detection result.

[0109] Preferably, performing cap and liquid level detection on each image in the fourth image group to obtain a cap defect detection result and a liquid level detection result includes:

[0110] S412-1. Perform image enhancement processing on each image in the fourth image group; by performing image enhancement processing on each image, the colors of the cap region and the liquid level region are deepened, and the contrast is improved. For example, the cap and the liquid become darker in black for subsequent detection.

[0111] S412-2. Based on a pre-trained positioning model, perform pattern positioning to obtain a pattern region; wherein, the pattern region includes a cap region and a liquid level region.

[0112] S412-3. Detect the contrast and the cap angle of the cap region to obtain the cap defect detection result; specifically, it includes the following steps:

[0113] Judge whether the contrast of the cap region is within a first preset interval. If the contrast is within the first preset interval, it is determined that the cap seal is qualified; otherwise, it is determined that the cap seal is unqualified to obtain a cap seal detection result; specifically, it includes: select a detection region in the cap region and correspondingly set the upper and lower limit intervals of the detection region to obtain the first preset interval. If the contrast of the cap region is within the interval, it is qualified; otherwise, it is unqualified. For example, judge whether all colors in the first preset interval are the color of the cap region (i.e., black). If so, it is determined that the cap seal is qualified; otherwise, it is determined that the cap seal is unqualified.

[0114] Judge whether the cap angle of the cap region is within a second preset interval. If the cap angle is within the third preset interval, the cap angle of the packaging container 7 to be tested is qualified; otherwise, the cap angle is unqualified to obtain a cap angle detection result; specifically, it includes: select two edge features for angle measurement and set the upper and lower limit intervals of the two edge features. If the cap angle is within the interval, it is qualified; otherwise, it is unqualified.

[0115] Obtain the cap defect detection result according to the cap seal detection result and the cap angle detection result.

[0116] S412-4. Detect the liquid level height of the liquid level area to obtain the liquid level detection result.

[0117] Judge whether the liquid level height of the liquid level area is within the third preset interval. If the liquid level height is within the first preset interval, the liquid level height of the to-be-tested packaging container 7 is qualified; otherwise, it is unqualified. Specifically, it includes: selecting two edge features, setting the upper and lower limit intervals of the liquid level. If the liquid level height is within the interval, it is qualified; otherwise, it is unqualified.

[0118] S413. If at least one of the label defect detection result, cap defect detection result, and liquid level detection result shows a defect, it is determined that the to-be-tested packaging container 7 has a local defect; otherwise, it is determined that there is no local defect.

[0119] S42. If the local defect detection result shows no defect, compare each image in the front image group and the top surface image with the corresponding preset standard image to determine whether there is a defect, and obtain the overall defect detection result.

[0120] Preferably, comparing each image in the front image group and the top surface image with the corresponding preset standard image to determine whether there is a defect includes:

[0121] S421. Preprocess each image in the front image group with the corresponding preset standard image. Among them, the preprocessing includes grayscale conversion, denoising, and normalization; use the Canny edge detection algorithm to extract the image edges; calculate the difference image and perform binary processing; count the total number of pixels in the defect area of the binary image. If the total number of pixels in the defect area is greater than the preset defect area threshold, it is determined that the front image group has a defect; otherwise, it is determined that the front image group has no defect.

[0122] In this embodiment, by counting the total number of pixels in the defect area of the binary image, it can be quickly determined whether there are appearance defects on the bottle body.

[0123] In one of the embodiments, it further includes dividing each image in the front image group into a pattern area, a text area, and other areas; comparing each area of each image with the corresponding area of the preset standard image. If any area has a defect, it is determined that the front image group has a defect; the other area is the background area where the pattern area and the text area are located.

[0124] In this embodiment, during the process of comparing each image in the front image group with a preset standard image, after preprocessing the images, each image in the front image group is segmented into three regions. During the comparison process, these three regions are respectively compared with the relevant regions in the preset standard pattern. When comparing each region of each image in the front image group with the corresponding region of each image in the preset standard image, wavelet transform is performed on each region of each image in the front image group and each region of the preset standard image to obtain wavelet decompositions with the same number of layers. The more layers, the more accurate the judgment effect. Thus, it is determined whether there are defects in the front image group.

[0125] S422. Convert the top surface image and the corresponding preset standard image into grayscale images; specifically, the weighted average method can be used to count the number of pixels of each gray level in the grayscale image; calculate the gray mean value, and truncate and amplify the gray difference of the defective area; among them, set a truncation range, and enhance the gray values that fall outside this truncation range to amplify the gray difference of the defective area. Use statistical methods to compare the gray histograms of the two images; compare the feature difference value with the threshold value to determine whether there are defects in the top surface image.

[0126] It should be noted that through the feature extraction method of the basic gray histogram, after gray conversion of the two RGB images to be compared, gray histogram statistics are performed, and statistical methods are used for comparison to obtain the feature difference value of the two images, and the threshold method is used to determine whether it is qualified. During the comparison process of the statistical method, the gray difference that may be caused by defects is truncated and amplified by using the gray mean value, which improves the accuracy of defect determination. High-speed quality determination in production line operation is realized.

[0127] S423. If at least one of the front image group and the top surface image has defects, it is determined that the packaging container 7 to be tested has defects; if not, it is determined that there are no defects.

[0128] S50. When the defect detection result is that there are no defects, obtain the previous defect information or the defect information within a preset time. The defect information includes defect type, defect image information, and defect detection method. Based on the defect type and the defect image information, determine the second target image with a suspicious defect area from the image group to be tested, adjust the detection parameters of the defect detection method, and use the adjusted detection parameters to perform defect detection on the suspicious defect area to obtain the final detection result; among them, the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before adjustment.

[0129] It should be noted that since the above defect detection result is the result of detecting the current packaging container 7 to be tested based on a preset defect detection strategy, and the detection parameters corresponding to the preset defect detection strategy are usually set based on the maximum limit under the qualified state. After a serious defect occurs, the corresponding equipment is usually repaired to improve the quality of the finished product. However, after the repair, the prior art usually does not consider the situation of the previous defect information or the defect information within a preset time. For example, perhaps the current repair result just meets the maximum limit under the qualified state, or there is a large instability and the repair is not actually perfect. In this case, if the detection result obtained by the preset defect detection strategy is still taken as the main result, the situation of insufficient repair is often ignored, and it is impossible to accurately know the result after repair based on the previous defect information or the defect information within a preset time. And this solution can actually solve this situation by adjusting the size of the detection parameters of the defect detection method and using the adjusted detection parameters to detect the suspicious defect area, thereby actually improving the product quality fundamentally. In addition, through this detection method, it is possible to accurately determine and precisely detect the occurred defects, so as to determine the corresponding repair accuracy. For example, within a specified time, the fewer defect results obtained by using the adjusted detection parameters to detect the suspicious defect area, the higher the repair accuracy, that is, the higher the relevance between this repair method and repair components and the defect problem, which is convenient for quickly and accurately determining the problem points and solution methods corresponding to the defects, continuously improving the quality of the finished product, and ultimately greatly improving the quality of the finished product in the entire production process.

[0130] In one embodiment, when the defect detection result is that there is no defect and the time length between the time stamp of the previous defect information or the defect information within a preset time and the current time stamp is greater than the preset time length, the defect detection result is used as the final detection result.

[0131] Please refer to Figure 5 , the present invention also provides a visual detection device for beverage packaging defects, and the device includes:

[0132] An acquisition module, configured to obtain an image containing the preset reference feature points of the packaging container 7 to be tested after detecting that the packaging container 7 to be tested moves to the detection imaging position, and obtain a first target image;

[0133] A determination module, configured to determine the rotation angle of the preset reference feature points according to the positions of the preset reference feature points in the first target image and the positions of the preset reference feature points in the standard target image;

[0134] A control module, configured to control the rotation of the camera group 4 based on the rotation angle of the preset reference feature points, and obtain an image of the corresponding plane of the to-be-detected packaging container 7 at the detection imaging position through the rotated camera group 4, so as to obtain a to-be-detected image group;

[0135] A first detection module, configured to perform defect detection according to the to-be-detected image group to obtain a defect detection result;

[0136] A second detection module, when the defect detection result indicates no defect, obtains the previous defect information or the defect information within a preset time. The defect information includes defect type, defect image information, and defect detection method. Based on the defect type and the defect image information, determine a second target image of the suspicious defect area from the to-be-detected image group, adjust the detection parameters of the defect detection method, and use the adjusted detection parameters to perform defect detection on the suspicious defect area to obtain a final detection result; wherein, the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before adjustment.

[0137] It can be understood that the functions or modules included in the system provided in this embodiment can be used to execute the methods described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments. For the sake of brevity, it will not be repeated here.

[0138] The present invention also provides an electronic device, including a processor and a memory. The memory is used to store computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes the method in any of the above possible implementation manners.

[0139] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by the processor of the electronic device, the processor is caused to execute the method in any of the above possible implementation manners.

[0140] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0141] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again. Those skilled in the art can also clearly understand that each embodiment of the present invention has different focuses in description. For the convenience and brevity of description, the same or similar parts may not be described in different embodiments. Therefore, the parts not described or not described in detail in a certain embodiment can refer to the descriptions in other embodiments.

[0142] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the devices or units can be in an electrical, mechanical, or other form.

[0143] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0144] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more integrated available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a digital versatile disc (DVD)), or a semiconductor medium (for example, a solid state disk (SSD)), etc.

[0145] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware with a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. The foregoing storage medium includes various media that can store program codes, such as read-only memory (ROM) or random access memory (RAM), magnetic disks, or optical discs.

Claims

1. A method for visually detecting beverage packaging defects, characterized in that: The method comprises: After detecting that the packaging container to be tested moves to the detection imaging position, acquiring an image including preset reference feature points of the packaging container to be tested to obtain a first target image; Determining a rotation angle of the preset reference feature point in the first target image according to the position of the preset reference feature point in the first target image and the position of the preset reference feature point in the standard target image; Controlling the rotation of the camera group based on the rotation angle of the preset reference feature point, and acquiring the image of the corresponding plane of the packaging container to be tested at the detection imaging position through the rotated camera group to obtain the image group to be tested; Perform defect detection according to the image group to be tested to obtain a defect detection result; When the defect detection result is that there is no defect, the last defect information or defect information within a preset time is obtained, the defect information includes defect type, defect image information and defect detection method, based on the defect type and the defect image information, a second target image with a suspected defect area is determined from the image group to be tested, the detection parameters of the defect detection method are resized, and defect detection is performed on the suspected defect area using the adjusted detection parameters to obtain a final detection result; wherein the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before the adjustment.

2. The method for visually detecting beverage packaging defects according to claim 1, characterized in that: Acquiring an image including preset reference feature points of the packaging container to be tested to obtain a first target image of the packaging container to be tested, including: The image of the corresponding plane of the packaging container to be tested at the detection imaging position is obtained by the camera group at the initial position to obtain a first image group; wherein the camera group includes four first image acquisition devices arranged in an array at the detection imaging position, and when the packaging container to be tested moves to the detection imaging position, the packaging container to be tested is at the focus position of the four first image acquisition devices; An image containing preset reference feature points is determined from the first image group, and the image containing the preset reference feature points is used as a first target image.

3. The visual detection method for beverage packaging defects according to claim 2, characterized in that: Determining a first target image including a preset reference feature point from the first image group includes: Using the first image acquisition device corresponding to the standard target image as the first target image acquisition device; Detecting the image captured by the first target image acquisition device to determine whether the preset reference feature point exists; If so, determining the image captured by the first target image capturing device as the first target image; If not, the remaining first image acquisition devices are randomly marked as the second target image acquisition device, the third target image acquisition device and the fourth target image acquisition device, and the images acquired by the second target image acquisition device, the third target image acquisition device and the fourth target image acquisition device are detected in sequence to determine whether the preset reference feature points exist, until an image with the preset reference feature points is detected, and the image with the preset reference feature points is determined as the first target image.

4. The visual inspection method for beverage packaging defects according to claim 1, characterized in that: Performing defect detection based on the image group to be tested to obtain defect detection results includes: The camera group includes a first image acquisition device and a second image acquisition device; wherein the first image acquisition device is used to obtain a front view of a corresponding surface of the packaging container to be tested, and the second image acquisition device is arranged directly above the detection imaging position, and the second image acquisition device is used to obtain a front view of a top surface of the packaging container to be tested to form a top surface image; The image group to be tested includes a front image group formed by front views of various planes of the packaging container to be tested, and the top surface image; Performing local defect detection on the front image group to obtain a local defect detection result; wherein the local defect detection includes one or more of label defect detection, cover defect detection and liquid level defect detection; If the local defect detection result is that there is no defect, each image in the front image group and the top surface image are compared with the corresponding preset standard image to determine whether there is a defect, thereby obtaining an overall defect detection result.

5. The visual inspection method for beverage packaging defects according to claim 4, characterized in that: Performing local defect detection on the front image group to obtain a local defect detection result includes: Acquire a region image containing a label region in each image of the front image group to obtain a third image group, generate a label expansion image based on the third image group, and perform defect detection on the label expansion image to obtain a label defect detection result; Acquire regional images including the cover area and the standard liquid level area in each image of the front image group to obtain a fourth image group, perform cover body and liquid level detection on each image in the fourth image group to obtain a cover body defect detection result and a liquid level detection result; If at least one of the label defect detection result, the cover defect detection result and the liquid level detection result indicates that there is a defect, it is determined that the packaging container to be tested has a local defect; otherwise, it is determined that there is no local defect.

6. The method for visually detecting beverage packaging defects according to claim 5, characterized in that: Generating a label expansion image based on the third image group and performing defect detection on the label expansion image includes: Preprocessing each image in the third image group; wherein the preprocessing includes smoothing filter processing and cylindrical label back-projection correction; splicing and fusing the preprocessed images in the third image group to obtain the label expansion image; The color information of the beverage and the label is used to determine whether there are label defects through double thresholds in the color space and a morphological processing algorithm.

7. The method for visually detecting beverage packaging defects according to claim 5, characterized in that: Performing cover body and liquid level detection on each image in the fourth image group to obtain a cover body defect detection result and a liquid level detection result includes: performing image enhancement processing on each image in the fourth image group; Based on the pre-trained positioning model, pattern positioning is performed to obtain a pattern area; wherein the pattern area includes a cover area and a liquid level area; Detecting the contrast and angle of the cover area to obtain the cover defect detection result; The liquid level height of the liquid level area is detected to obtain the liquid level detection result.

8. The method for visually detecting beverage packaging defects according to claim 4, characterized in that: Comparing each image in the front image group and the top image with the corresponding preset standard image to determine whether there is a defect, including: Preprocessing each image in the front image group and the corresponding preset standard image, wherein the preprocessing includes graying, denoising and normalization; extracting image edges using the Canny edge detection algorithm; calculating a differential image and performing binarization processing; counting the total number of pixels in the defective area in the binarized image; if the total number of pixels in the defective area is greater than a preset defective area threshold, determining that the front image group has defects, otherwise determining that the front image group does not have defects; Convert the top surface image and the corresponding preset standard image into grayscale images; count the number of pixels at each grayscale level in the grayscale image; calculate the grayscale mean, and truncate and amplify the grayscale difference of the defect area; use a statistical method to compare the grayscale histograms of the two images; and determine whether the top surface image has defects based on the comparison of the feature difference value with the threshold value; If at least one of the front surface image group and the top surface image shows a defect, it is determined that the packaging container to be tested has a defect; if not, it is determined that there is no defect.

9. A visual inspection device for beverage packaging defects, characterized in that: The device comprises: An acquisition module, configured to acquire an image including preset reference feature points of the packaging container to be tested, and obtain a first target image after detecting that the packaging container to be tested moves to the detection imaging position; a determination module, configured to determine a rotation angle of a preset reference feature point in the first target image according to a position of the preset reference feature point in the first target image and a position of the preset reference feature point in the standard target image; A control module, used for controlling the rotation of the camera group based on the rotation angle of the preset reference feature point, and obtaining the image of the corresponding plane of the packaging container to be tested at the detection imaging position through the rotated camera group to obtain the image group to be tested; A first detection module, used for performing defect detection according to the image group to be tested to obtain a defect detection result; The second detection module obtains the last defect information or the defect information within a preset time when the defect detection result is that there is no defect, the defect information includes the defect type, defect image information and the defect detection method, determines a second target image with a suspected defect area from the image group to be tested based on the defect type and the defect image information, adjusts the detection parameters of the defect detection method, and performs defect detection on the suspected defect area using the adjusted detection parameters to obtain a final detection result; wherein the detection accuracy corresponding to the adjusted detection parameters is higher than the detection accuracy corresponding to the detection parameters before the adjustment.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the visual detection method for beverage packaging defects according to any one of claims 1 to 8.

Images