How to detect defects in horizontal mold seals for glass containers

Abstract

The present invention relates to a method for detecting defects in a horizontal mold seal (JH) for a finish (B) of a plurality of glass containers (R), the method comprising the following steps: placing the container (R) between a light source and a camera, ensuring rotation of the container (R) for one rotation, acquiring an image by the camera at each container rotation increment such that the number of images per rotation is greater than 36, and analyzing the captured images for each container, the images being captured such that a contour of the end of the finish is detected in each image, comparing the contour of the end of the finish in the image with a reference contour of the end of the finish to detect deviations between these contours, and a defect in the horizontal mold seal (JH) for a container is detected if at least one image of the container has a deviation.

Description

[Technical Field]

[0001] The present invention relates to the technical field of inspection of objects, hollow articles or generally transparent or translucent containers such as, for example, glass bottles, jars or flasks.

[0002] The object of the present invention is more particularly to provide a horizontal mold for finishing such glass containers. seam The present invention relates to the field of inspection of glass containers for the purpose of detecting the presence or absence of defects. [Background technology]

[0003] Generally, a container has a bottom from which a vertical wall rises, terminating in a section called a finishing section. There are various types of finishing sections depending on the closure system installed. As shown in Figure 1, finishing sections include an annular finishing surface S at the top and an unfinished section CB at the bottom. Figure 1 shows a finishing section that is threaded and includes a thread on the cylindrical vertical section. Glass containers are known to be manufactured by a forming machine called an IS machine, which includes various independent forming sections to which drops of malleable glass are fed. Each of these forming sections is equipped with at least one blanking cavity with a blank mold and an equal number of finishing cavities, each of which has a blow mold in which the container is given its final shape at high temperature.

[0004] Conventionally, the finishing section is formed in the blank mold. When the blank is transferred to the blow mold, the finishing section is already formed and the blank is held by the finishing section. To this end, the finishing section is made up of two half molds that form the vertical walls of the finishing section and the surface or the part of the finishing section. seamThe blank mold also contains two blank body half molds to form the walls of the blank body. During the press-and-blow process, the punch presses the glass against the blank body half molds. During the blow-and-blow process, the punch shortens and enters the finishing mold, and compressed air presses the glass against the blank body half molds. Therefore, the outside of the vertical walls of the finishing section are formed by the two finishing half molds, the top surface of the finishing section is formed by the ring, and the inside of the finishing section is formed by the punch. As a result, the finishing section of each container R has vertical molds on both sides of its vertical walls that correspond to the interface between the two finishing half molds, as shown in Figure 1. seam The horizontal die corresponds to the interface between the JV and the finishing die section called the ring and the two finishing half dies. seam These finishing dies include JH and seam The markings on the container will be more or less pronounced depending on the adjustment of the mold parts that deteriorate with use.

[0005] Horizontal mold seam It should be taken into consideration that JH is located slightly below the surface S of the finished part of the container. seam It appears necessary to detect defects affecting the aesthetic qualities of the container or, more seriously, to eliminate containers containing defects that may pose a real danger to the user.

[0006] In the latest technology, AGR INTERNATIONAL uses DSG to accurately measure the dimensional properties of containers. 40 laboratory monitoring machines, horizontal molds known as knockouts and flanges seam This machine in particular comprises a light source arranged on one side of the container and a camera arranged on the other side of the container. When capturing the images, the container is driven in rotation about its vertical axis. Such a machine is suitable for horizontal molds. seamIt appears that the machine is not designed to detect all defects in the container. Furthermore, the monitoring rate of this machine is limited, so it is not possible to monitor containers at the known production rate on a container production line.

[0007] WO 2013 / 128538 describes a horizontal mold seam A method for detecting defects in the finish of glass containers is described. The method involves projecting a beam of light perpendicularly onto the finish while the container itself is rotated. A linear camera collects the light reflected by the finish. The method involves analyzing the reflected light to infer defects when the profile of the reflected light changes. While this method has a very high level of reliability for detecting small defects, it can only detect flanges that reflect light in the direction of the camera, cannot quantify the size of the defect, and is sensitive to the horizontality of the container mold. seam It is not possible to detect various types of defects in the

[0008] Thus, the container mould can be levelled at high rates while having a very high level of reliability for detecting small, typically less than 1 mm, defects. seam While various defects in the mold of such containers can be detected, seam There appears to be a need for technology that can avoid considering defective containers if they actually meet the required quality standards. Summary of the Invention [Problem to be solved by the invention]

[0009] Therefore, an object of the present invention is to provide a horizontal mold for finishing glass containers at a high rate. seam The present invention aims to meet this need by proposing a method for detecting defects in the horizontality of the mold of such containers. seam horizontal molds while avoiding the consideration of defective containers if they actually meet the required quality standards seamIt has a very high level of reliability for detecting small defects in [Means for solving the problem]

[0010] To achieve this object, the present invention provides a horizontal mold for finishing a plurality of glass containers, each having a vertical axis. seam The present invention proposes a method for detecting defects in a container, the method for detecting defects for each container comprising the steps of: placing the container between the light source and the image capture camera, wherein the observation optical axis of the image capture camera is substantially perpendicular to an axis parallel to the vertical axis of the container, and the field of view of the image capture camera is horizontal; seam At least a part of the left end of the finishing part or the right end of the finishing part, ensuring rotation of the container along its vertical axis through at least one revolution; acquiring an image by the camera at each rotational increment of the container such that the number of images per rotation is greater than 36; and analyzing the captured images for each vessel; Including, The image above is Horizontal mold seam a finish inspection zone spanning the height of the finish including at least a portion of: The contour of the edge of the finish is detected in the image inspection zone. comparing the finish edge contours of the image with reference finish edge contours to detect deviations between these finish edge contours and the reference finish edge contours; and Horizontal mold for containers seam a defect in the container is detected when at least one image of the container has a deviation, It has been captured.

[0011] According to an advantageous variant of embodiment, the camera has a field of view of a horizontal type. seam At least part of the horizontal mold seamand a reference whose elevation position is known relative to the horizontal mold, and which defines a finish inspection zone in the image; seam A reference whose elevation position is known relative to the horizontal mold is identified, and the finishing inspection zone is seam The axial length of the slit is positioned relative to a reference so as to extend over a height including the axial length of the slit.

[0012] For example, all or part of a finished or unfinished surface may be identified in the image as a reference.

[0013] Typically, there are image capture cameras with a horizontal field of view width of 5mm to 130mm and a field of view height of 3mm to 20mm.

[0014] For example, there are image capture cameras that have resolutions greater than 25 pixels / mm.

[0015] According to one preferred feature of implementation, the image capture camera is positioned so that its observation optical axis is substantially tangent to the left or right edge of the finish.

[0016] The method according to the invention can use a camera equipped with a telecentric lens.

[0017] According to one advantageous feature of the embodiment, there is a light source having vertical and horizontal illumination dimensions of 100% to 200%, preferably 100% to 120%, of the dimension of the field of view of the camera multiplied by the distance between the light source and the camera lens and divided by the distance between the axis of the container and the camera lens.

[0018] According to another embodiment, there is a telecentric light source, the illuminated area of ​​which has dimensions equal to or greater than the dimensions of the field of view of the camera.

[0019] According to one variant of the embodiment, the contour of the end of the finish in the image is compared with a reference contour of the end of the finish by using as reference contour at least one contour obtained for at least one reference container that is considered to be suitable, and deviations are detected.

[0020] According to another variant of the embodiment, the contour of the end of the finish in the image is compared with a reference contour of the end of the finish by using as the reference contour an average contour calculated over several rotational increment values ​​of at least one reference container that is considered to be suitable, and deviations are detected.

[0021] According to another variant of the embodiment, the contour of the end of the finish in the image is compared with a reference contour of the end of the finish by using the low-pass filtered contour of the end of the finish as the reference contour, and deviations are detected.

[0022] According to another variant of embodiment, the contour of the edge of the finish in the image is compared with a reference contour of the edge of the finish by applying a high pass filter to the contour of the edge of the finish and deviations are detected.

[0023] Advantageously, for each image, the contour of the end of the finish and the reference contour of the end of the finish are compared for at least some elevations in the inspection zone by comparing at least one of the area, amplitude and / or slope measurements with a threshold value, and a deviation is detected if at least one of these measurements exceeds this threshold value.

[0024] Generally, the method includes: detecting at least one horizontal mold selected from the group consisting of knockout, flange, and overhang / overmatch defects; seam This includes detecting defects.

[0025] According to one preferred embodiment, at least one detection criterion is defined for each of the following defects: knockout, flange, and overhang / overmatch, and the method includes identifying the detected defect from among these three defects by using at least one of these detection criteria.

[0026] According to one feature of implementation, the shape of the deviations detected for several heights in the examination zone, for at least one image, is selected as detection criterion.

[0027] According to another feature of implementation, the observation angle range of the deviation corresponding to the number of consecutive images in which the deviation is detected is selected as the detection criterion.

[0028] The shape of the detected deviations for several elevations in the inspection zone is analyzed to distinguish overhang / overmatch defects from knockout and flange defects.

[0029] To distinguish knockout defects from flange defects, the deviation observation angle range is analyzed by considering a knockout defect to be detected if the deviation observation angle range is below a certain maximum value.

[0030] To distinguish knockout defects from flange defects, the deviation observation angle range is analyzed by considering a flange defect to be detected if the deviation observation angle range is above a certain minimum value.

[0031] Conventionally, horizontal mold seam For each container that has a defect, a signal is sent to eject that container from the production line.

[0032] According to one advantageous feature of implementation, the container rotates along its vertical axis to make at least one revolution during a maximum duration of 200 milliseconds.

[0033] Another object of the present invention is to provide a horizontal mold for finishing glass containers. seam The object of the present invention is to propose an inspection device for implementing the method according to the invention for detecting defects in a semiconductor device. [Brief explanation of the drawings]

[0034] [Figure 1] FIG. 1 is an enlarged view showing the finish, horizontal mold seam, and each vertical mold seam of a glass container.

[0035] [Figure 2] FIG. 2 is a schematic elevational view showing an apparatus for carrying out the method according to the invention for detecting horizontal mold seam defects on glass container finishes.

[0036] [Figure 3A] FIG. 3A is a side view of the end of a finished portion of a glass container having a knockout-type horizontal mold seam defect. [Figure 3B] FIG. 3B is a top view of a finished portion of a glass container having a knockout-type horizontal mold seam defect.

[0037] [Figure 4A] FIG. 4A is a side view of the end of a finished portion of a glass container having a flange-type horizontal mold seam defect. [Figure 4B] FIG. 4B is a top view of a finished portion of a glass container having a flange-type horizontal mold seam defect.

[0038] [Figure 5A] FIG. 5A is a side view of the end of a finished portion of a glass container having an overhang / overmatch type horizontal mold seam defect. [Figure 5B] FIG. 5B is a top view of the finished portion of a glass container having an overhang / overmatch type horizontal mold seam defect.

[0039] [Figure 6] FIG. 6 is an example image of the right edge of a finish on a glass container with a defect in the horizontal mold seam.

[0040] [Figure 7] FIG. 7 illustrates the extraction in an image of the contour of the edge of the finish of a container having a defect in the horizontal mold seam.

[0041] [Figure 8] FIG. 8 is a diagram illustrating the extraction in an image of the reference contour of the end of the finishing part from the matching container.

[0042] [Figure 9] FIG. 9 is a diagram illustrating a comparison of the contour of the end of the finish of a container having a flange or knockout-type horizontal mold seam defect of the container with the reference contour of the end of the finish.

[0043] [Figure 10] FIG. 10 is a diagram illustrating a comparison of the contour of the edge of the finish of a container having an overhang / overmatch type horizontal mold seam defect with the reference contour of the edge of the finish.

[0044] [Figure 11] FIG. 11 is a diagram illustrating extraction of the reference contour of the end of the linearly shaped threaded finishing portion from an image.

[0045] [Figure 12] FIG. 12 is an example image of the right end of a finish on a container having a knockout defect, showing the container outline and the reference outline. DETAILED DESCRIPTION OF THE INVENTION

[0046] As can be more particularly seen from FIGS. 1 and 2, the object of the present invention is to provide a horizontal mold for glass container R. seamIt relates to an apparatus 1 for carrying out the method according to the invention for detecting JH defects. Conventionally, a container R has a bottom F from which rises a vertical cylindrical wall V along a vertical axis Z, terminating in a section called the finish B. In the case of a bottle-type container R, the vertical cylindrical wall V emerges from the bottom F, which is the part that forms the body of the bottle, which is connected to a collar C via a shoulder E.

[0047] Each container R is formed by a horizontal mold section, as shown in FIG. 1, which corresponds in particular to the interface between the mold section called the ring and the two mold halves that form the vertical cylindrical wall V of the container. seam Including JH. By definition, horizontal mold seam JH is a container seam The horizontal die is located slightly below the surface S of the finishing part, which corresponds to the plane for finishing. seam The JH extends around the entire circumference of the finish.

[0048] According to the present invention, the method is for detecting defects in horizontal molds selected from the group consisting of knockout JHK, flange JHF, and overhang / overmatch JHO defects. seam This includes detecting defects in the JH for each container.

[0049] As can be seen more specifically from Figures 3A and 3B, the knockout JHK defect is a horizontal mold seam The vertical mold between the JH and the two half molds that form the vertical wall and the mold section called the ring. seam This corresponds to a glass flange at one of the T-shaped intersections of the JHK and the JV. This defect corresponds to a glass flange caused by a blunt bulge at the interface of the three molds or due to misalignment between these molds. This knockout JHK takes the form of a chip- or needle-like glass protrusion (often triangular in cross section) with a thin thickness along the vertical axis Z and a small horizontal extent, often less than 1 millimeter. As a result, it has a small circumferential angular extent θ about the vertical axis Z. Typically, the knockout JHK has a small circumferential angular extent θ of less than a few degrees.

[0050] As can be seen more specifically from Figures 4A and 4B, a flange JHF defect corresponds to excess glass resulting from misalignment between the mold section known as the ring and the two mold halves that form the vertical wall. This flange JHF takes the form of a thin glass slide aligned along a vertical axis Z. Furthermore, this flange JHF has a significant angular extent θ about the vertical axis Z. Typically, the flange JHF has a significant angular extent θ of 10° or more, but easily exceeding 30°. It can be even wider, up to 180°. In rare cases, the flange may be present around the entire circumference.

[0051] As can be seen more clearly from Figures 5A and 5B, the overhang / overmatch JHO defect corresponds to excess glass resulting from misalignment between a mold section called the ring and the two mold halves that form the vertical wall. This overhang / overmatch JHO takes the form of a very thick glass protrusion along a vertical axis Z and has a significant angular range θ about the vertical axis Z. Typically, the ring offset can appear to exceed 180°, but the resulting step is generally noticeable over an angular range θ of 90° or more.

[0052] As will be described in detail below, the object of the present invention is to provide a horizontal mold for preventing or reducing defects selected from any or all of the following: knockout JHK, flange JHF, and overhang / overmatch JHO defects. seam According to an advantageous variant of embodiment, the object of the invention is to detect containers having defects in the JH. According to an advantageous variant of embodiment, the object of the invention is to detect horizontal mold defects selected from among knockout JHK, flange JHF, and overhang / overmatch JHO defects. seam The aim is to identify defects in JH.

[0053] For this purpose, horizontal molds seam The device 1 for carrying out the method for detecting defects in a JH comprises a stationary light source 3 arranged on one side of the container R and a stationary image capturing camera 4 arranged on the other side of the container R. This camera 4, including a lens 4a, is adapted to capture an image of the horizontal mold of the container R. seamThe camera 4 is adapted to capture an image in which at least a portion of the JH is visible. The camera 4 analyzes the captured image and determines the horizontal orientation of the container R. seam The JH is connected to an analytical processing unit 5 configured to detect defects in the JH.

[0054] The images of the container R are captured while the container R rotates at least once around its vertical axis Z, thereby forming a horizontal mold. seam The entire JH can move in front of the camera. Each container R is supported by a rotation system 6, which for example includes a sliding or laying surface 7 for the bottom F of the container R, as well as a drive system 8 of any type known per se. The drive system 8 is operated so that each container R remains between the light source 3 and the camera 4 for the time required for at least one rotation, during which an image is captured by the camera 4, as will be explained below.

[0055] For example, the rotation system 6 includes, as a drive system, a wheel or spinner 8 that drives the container R in rotation by friction, while the container is supported on at least two free casters or rollers 9. The rollers, for example, form part of a transport starwheel 10 that transports the container R on a circular path and continuously supports and slides the container R on the laying surface 7 in front of the detection device 1. Advantageously, this transport starwheel 10 forms part of an inspection machine that inspects more than 150 containers per minute and includes one or more stations for inspecting containers in motion at the exit of the production line. In other words, the detection device 1 according to the invention can be implemented in addition to inspection stations customarily installed to inspect containers on the production line, without reducing the container run rate at these stations that integrate the rotation of the containers for this inspection.

[0056] According to a feature of the present invention, the in-line inspection machine inspects 50 to 500 containers per minute, typically 150 to 450 containers per minute. For each container, the method includes a transport step, which involves transporting the container R and positioning it between a light source 3 and an image capture camera 4, and a step of ensuring rotation of the container R along a vertical axis Z for at least one revolution, during which images are acquired. Note that, using the transport starwheel 10, a previously inspected container is brought to the inspection station at the same time as a new container to be inspected is introduced. Thus, the in-line inspection machine is equipped with a high-rate handling means. At a rate of 150 containers per minute, the duration of the rotation and inspection step is a maximum of 200 milliseconds. According to one preferred feature of the present invention, a complete rotation of the container lasts a maximum of 100 milliseconds. Rotating the container 1.5 times is also frequently selected.

[0057] In other words, the method according to the present invention comprises placing the container R between the light source 3 and the image capture camera 4, wherein the observation optical axis of the image capture camera 4 is substantially perpendicular to an axis parallel to the vertical axis Z of the container, and the field of view of the image capture camera 4 is substantially perpendicular to an axis parallel to the vertical axis Z of the container, and seam and ensuring rotation of the container R along the vertical axis Z to complete at least one rotation in less than 200 milliseconds, the rotation including at least a portion of the left end of the finishing section or the right end of the finishing section, and capturing images by the camera 4 at each rotation increment of the container such that the number of images per rotation is greater than 36.

[0058] According to one aspect of the present invention, the camera 4 is operated so that an image is captured at each container rotation increment such that the number of images per rotation is greater than 36. In other words, the method of the present invention aims to acquire at least one image for every 10° of container R rotation. For example, the number of images over 360° of container R may be between 36 and 96. The container rotation increments between each captured image represent the angular sector through which the finishing part passes, for example, from 10° to less than 1°. Of course, a high-speed camera with an integration time of less than 500 microseconds and a read time of 0.5 milliseconds can acquire 400 images per rotation in 200 milliseconds. Increasing the acquisition frequency leads to improved accuracy, especially in estimating the radial length of the knockout. As a result, the cost of the system is high due to the price of the high-speed camera 4 and the computing power that the analysis and processing unit 5 must possess.

[0059] According to the invention, the light source 3 and the camera 4 are arranged to detect the horizontal shape of the container R during the rotation of the container about the vertical axis Z. seam The camera is adapted to acquire an image I in which at least a portion of JH appears. As shown in Figure 2, by convention, the vertical axis Z of the container is considered to be parallel to the vertical direction z of an orthogonal reference frame x, y, z. This vertical direction z is perpendicular to the transverse direction y passing through the light source 3 and the camera 4, while the lateral direction x is perpendicular to the transverse direction y and the vertical direction z.

[0060] The container R is located between the light source 3 and the camera 4, whose observation optical axis Y extends parallel to the transverse direction y, i.e., substantially perpendicular to the vertical direction z. The camera 4 has a field of view extending laterally along a lateral direction x perpendicular to the vertical direction z and the observation optical axis Y. The field of view of the camera 4 therefore extends within a plane defined by the vertical direction z and the lateral direction x. According to a preferred example, the camera 4 is positioned so that its optical axis is substantially tangent to the left or right edge of the finish.

[0061] In accordance with the vertical direction z, the field of view of the camera 4 includes at least one horizontal die. seamJH. The field of view of camera 4 includes at least the left end of finishing section B or the right end of finishing section B. According to one preferred implementation variant, the field of view of camera 4 includes only either the left end of finishing section B or the right end of finishing section B in order to obtain good resolution. Of course, the method according to the invention can be carried out using a camera whose field of view is not limited to the right end or the left end. This preferred variant is achieved when the in-line inspection machine rotates the container at least once, and the observation of one side is performed by observing the horizontal mold of the finishing section. seam It is particularly suitable for high rates, as it has been found to be sufficient to ensure the observation of defects in the JH. Furthermore, single-sided observation makes it possible to optimally utilize the camera's field of view and to position its optical axis tangent to the edge of the finish being observed.

[0062] This camera 4, like a matrix camera, transmits for each rotation increment a horizontal projection of the finish to the analysis and processing unit 5, in which at least one edge of the finish, called the left edge or the right edge, stands out, taking into account their position appearing in each image captured by the camera (Fig. 6). The left or right term is given from the viewpoint of observation in all images, since a rotating container does not, strictly speaking, include either the left or the right side. In the example displayed in Fig. 6, image I corresponds to the right edge of container R.

[0063] Therefore, at least the left end of the finishing part B or the right end of the finishing part B appears in each image captured by the camera, and thus the horizontal mold seam It is possible to visualize at least a part of the JH and at least the contour P of the end of the finishing part. In other words, as can be seen in FIG. 6, each image I is a horizontal mold at an elevation or height along a direction parallel to the vertical direction z. seam It must include at least a portion of the contour or outline P of the end of the finished part, including at least a portion of JH. seam An inspection zone ZI is defined which corresponds to a horizontal band of the image having a height determined to include part of the finish edge contour P, including part of JH.

[0064] According to an advantageous feature of the implementation, the camera 4 has a horizontal field of view. seam At least part of the JH and horizontal mold seam The JH is positioned to include a reference whose altitude position is known. Therefore, after finding the reference in the image, the finishing inspection zone can be sure that the horizontal mold seam It is possible to position the finish inspection zone ZI relative to the reference so that it extends over a height including

[0065] Of course, this image capture is performed in accordance with this reference and horizontal mold, taking into account variations in the positioning of the container R relative to the camera 4 and manufacturing tolerances of the container R. seam This is performed with an additional margin to allow visualization of part of the JH. If the vessel height can vary by + / - 2 mm, this margin is, for example, at least + / - 2 mm.

[0066] It should be understood that this criterion presented by the container R is a notable element, such as a contour, angle, shape, etc., that can be shown in the image. This notable element can also be seen in all images captured during the rotation of the article, and its vertical position in each image can be uniquely determined, and the horizontal mold seam Its perpendicular distance Dv to JH is known in each image, so this distance is preferably constant. The location of this fiducial in the image allows each image to be assigned a positional reference frame, which can be used in analyzing the images.

[0067] Typically, for example, all or part of the surface S of the finished portion or the surface of the unfinished portion CB, if the container includes it, can be taken as a reference. When the surface S of the finished portion is taken as a reference, the camera 4 is adjusted so that its height of view includes the surface S of the finished portion plus a margin to account for variations in container positioning and manufacturing tolerances. Similarly, the camera 4 is adjusted so that its height of view includes the surface S of the finished portion plus a margin to account for variations in container positioning and manufacturing tolerances. seam Adjusted to include JH plus margin.

[0068] The horizontal width of the field of view of camera 4 is adapted to include the contour P of the edge of the finish with the margin and the reference, i.e., in the considered example, a part of the surface of the finish. According to one preferred variant, the width of the field of view is selected to include only the edge of the finish plus the necessary margin. For example, camera 4 preferably has a horizontal field of view width of 5 mm to 80 mm and a field of view height of 3 mm to 20 mm. This makes it possible to optimize the field of view of the camera, i.e., to limit the unnecessary area of ​​the sensor. According to another variant, the field of view includes both the left and right edges of the finish, and the field of view width can reach 130 mm to observe a finish with a diameter of 120 mm and a margin of 10 mm.

[0069] For example, the camera 4 is a matrix camera combined with an optical lens, making it possible to observe the inspection zone with an overall resolution of more than 25 pixels / mm.

[0070] The light source 3 is made in any suitable way to ensure backlighting of the container R adapted to capture an image by the camera 4 .

[0071] According to one preferred feature of the embodiment, the light source 3 has determined vertical and horizontal illumination dimensions. Conventionally, the vertical dimension is taken along the vertical direction z, and the horizontal dimension is parallel to the lateral direction x. In a non-telecentric optical system, the dimension DL of the light source 3 is of the following type: DL = CH × DI / DC; where CH is the field of view of the camera 4, DI is the distance between the light source 3 and the camera lens 4a, and DC is the distance between the inspected zone and the camera lens. More specifically, DC is considered as the distance between the camera lens (e.g., the optical center of a non-telecentric lens) and the focal plane, which includes the vertical axis Z of the container and the point of contact of the light ray passing through the optical center with the edge of the finish. The vertical and horizontal illumination dimensions are the dimensions of the effective area of ​​the light source 3 that can be observed by the camera 4.

[0072] According to one feature of the embodiment, the light source 3 has vertical and horizontal illumination dimensions of 100% to 200%, preferably 100% to 120%, of the field of view of the camera 4 multiplied by the distance between the light source 3 and the camera lens 4a and divided by the distance between the vertical axis Z of the container and the camera lens 4a.

[0073] According to one variant of embodiment, the light source 3 can be a telecentric light source whose illumination area has dimensions equal to or greater than the dimensions of the field of view of the camera 4. Thus, for the field of view of the camera, the light source 3 generates a beam whose average rays are parallel to the optical axis of the camera. It should be noted that in this case the lens 4a of the camera is a telecentric lens.

[0074] The image captured by the camera 4 for each container R shows the horizontal shape of the container R. seam The horizontal mold is analyzed by an analysis processing unit 5 configured to detect defects in the JH. seam The method for detecting defects in the JH involves, for each container, determining the captured images and, in particular, the horizontal mold seam and analyzing a finish inspection zone ZI spanning the height of the finish including at least a portion of the height of the container. The analysis method includes detecting contours P of the edges of the finish in the image inspection zone, and comparing the contours P of the edges of the finish in the image with a reference contour Pf of the edges of the finish to detect deviations between these contours P of the edges of the finish and the reference contour Pf of the edges of the finish. seam A defect in is detected if at least one image of the container R has a deviation. seam For each container with a JH defect, the analytical processing unit 5 sends a signal indicating the poor quality of the container, and such a signal can control an ejection device to eject the container from the production line.

[0075] The detection of the finish edge contour P in each captured image can be performed by any suitable image processing method.

[0076] As explained above, the analysis of the images is preferably limited to the finishing inspection zone ZI, where the horizontal mold seam Advantageously, the inspection zone is positioned relative to the reference of the container appearing in the image, and its elevation position is such that the horizontal mold seam In a variant of this implementation, the analysis of the contour of the finished part is carried out using a horizontal die. seam While the horizontal mold is limited to the parts necessary to characterize the seam The present invention excludes other contours that may be confused with the contours of the periphery, such as contours containing threads.

[0077] According to this method, horizontal mold seam A reference whose elevation position is known relative to the reference is identified in each image, and the finishing inspection zone is always aligned with the horizontal mold. seam It is positioned so as to extend over a height including JH.

[0078] FIG. 7 shows, as an example, the detection of the contour P of the right edge of the finish in an image I, where the surface S of the finish appears as a reference. The first step involves locating the height, i.e., the location of the surface S of the finish along the vertical direction z. For example, along a vertical line following the left edge of the image, a vertical gray level transition corresponding to the surface S of the finish is detected. The image is then inspected for horizontal mold defects. seam From this reference the inspection zone ZI can be positioned so that JH is necessarily taken into account.

[0079] Of course, the method of searching for the reference can be different: for example, if the reference is a non-finish, a search for the maximum two-dimensional correlation can be performed between the image and a reference window containing a trained model of the non-finish or of the visible right or left edge of the non-finish.

[0080] The second step is, for example, from the surface S of the finishing part to a horizontal mold below the surface of the finishing part. seamThe process includes extracting the outer contour or profile P of the end of the finishing part to the lower limit of the inspection zone ZI, which is located beyond ZH. As shown in FIG. 7, the contour P of the end of the finishing part is seam JH. The search for the first contour point P of the finish edge may involve scanning the image horizontally at a predetermined height relative to the reference until a horizontal grey level transition is detected.

[0081] Extraction of the contour P of the edge of the finish in each image can be performed using any image processing method. For example, it can be provided in the inspection zone ZI to search for the outer edge of the finish by searching for a black / white transition Ti for each level along the vertical direction z, or by searching for a white / black transition from the outside. The position of this transition Ti along the lateral direction x is determined relative to the vertical direction z. The search for this transition Ti is performed at each level throughout the entire height of the inspection zone ZI. As shown on the right of Figure 7, for each image, the contour P of the edge of the finish can be obtained in the form of a curve referenced in the plane z, x. For example, the contour P is made up of all the transitions Ti, the coordinates x and z of each transition being known.

[0082] The next step involves comparing the contour P of the end of the finishing part with a reference contour Pf of the end of the finishing part, obtained by various possible means. According to one preferred variant of the invention, as can be seen in Figure 8, the extraction of the reference contour Pf from the end of the finishing part is carried out in the same way as the extraction of the contour P of the end of the finishing part of the container shown in Figure 7. For this purpose, a defect-free horizontal reference mold is used. seam When an image is captured of a reference container including a reference contour Pf of the end of the finish, including part of the surface of the finish, it is possible to visualize this reference contour Pf of the end of the finish and, as a reference, the surface of the finish. This reference image is advantageously created under the same conditions as the image of the container to be inspected using the inspection device by placing the reference container between the camera 4 and the light source 3. By analyzing this image in the manner described above, it is possible to determine whether a defect-free horizontal mold is present. seam It is possible to extract the reference contour Pf of the end of the finished part where

[0083] According to another embodiment variant, the reference contour Pf of the end of the finisher is obtained by learning from several images of the container or from several images of containers that are considered to be compatible. For example, any type of mathematical operation can be performed on the images or on the contours extracted from the images. Thus, one or more contours obtained for the reference container that is considered to be compatible or for several reference containers that are considered to be compatible can be selected as the reference contour Pf. Similarly, an average contour calculated over several rotation increments of the reference container that is considered to be compatible or for several reference containers that are considered to be compatible can be selected as the reference contour Pf. According to another embodiment variant, the reference contour Pf of the end of the finisher can also be obtained from a manufacturing drawing or from any geometric model of the container.

[0084] According to another variant of the invention shown in Figure 11, for example in the case of the threaded finisher shown in Figure 1, the end of the finisher includes a cylindrical portion, so that the profile of the end of the finisher includes a right portion PD. In this case, the reference profile Pf of the end of the finisher can best be a simple reference straight line D that passes through the straight line portion PD of the profile P of the end of the finisher, and thus passes through a point Ti of this portion of the outline of the finisher, or a straight line parallel to it.

[0085] According to another variant of the embodiment shown in Fig. 12, the present invention aims to use a low-pass filtered contour of the end of the finish as the reference contour Pf. In other words, the reference contour Pf corresponds to a filtered or smoothed contour of the end of the finish. As an example in Fig. 12, the finish end contour P (white) corresponds to the raw contour extracted from the image, and the reference contour Pf (gray) corresponds to a moving average of the contour P. Of course, various types of low-pass filters can be applied, such as an averaging filter, a Gaussian filter, a median filter, or a phase-shift median filter.

[0086] It should be noted that, according to another variant of the embodiment, it may be envisaged to apply a high-pass filter to the finish edge contour P of the image in order to compare it with the finish edge reference contour Pf. The application of a high-pass filter, for example of the gradient type, makes it possible to highlight deviations, i.e. strong or rapid variations in the derivative of the finish edge contour, which reflect the presence of local irregularities. The presence of such deviations can be detected by the horizontal die. seam This corresponds to a defect in

[0087] A comparison step for each image between the contours P of the end of the finish in the image and the reference contour Pf of the end of the finish leads to detecting whether there are any deviations between these contours of the end of the finish and the reference contour of the end of the finish. seam A defect in is detected if at least one image of the container has a deviation.

[0088] Horizontal mold seam This comparison, which leads to the observation of the presence or absence of defects, can be carried out according to various analytical methods.

[0089] For each image, for at least some elevations, preferably for all elevations in the inspection zone ZI, the finish edge contour P is compared with the finish edge reference contour Pf. This step aims to compare at least one of the area, amplitude and / or gradient measures with a threshold value, and a deviation is detected if at least one of these measures exceeds this threshold. Typically, the area measure may correspond to the area between the finish edge contour P and the finish edge reference contour Pf. The amplitude measure may correspond to the difference between the finish edge contour P and the finish edge reference contour Pf along the lateral direction x. The gradient measure is obtained by subtraction of the finish edge contour P from the finish edge reference contour Pf. Another possible measurement is the difference between the amplitudes of the end of finish profile P and the reference end of finish profile Pf, at the same or different altitudes.

[0090] According to one variant of the embodiment, to compare the finishing end contour P with the finishing end reference contour Pf, the actual position of the finishing end is taken into account, including the tilt in the image related to the tilt of the finishing end relative to the Z axis due to operational disturbances during monitoring or lack of verticality of the container. Such a finishing tilt is shown, for example, in FIG. 11. It should be noted that if the finishing end reference contour Pf is a straight line D passing through the linear portion PD of the finishing end contour, in this simplified variant the deviation can be determined by measuring the distance between several points of the finishing contour P and the reference line D.

[0091] In the example shown in FIG. 9, the curve CEc corresponds to the deviation between the finish edge contour P and the finish edge reference contour Pf, which gradually evolves along the vertical axis z, i.e., depending on the height in the inspection zone. Various possible measurement examples of the deviation curve CEc make it possible to estimate the size and / or shape of the deviation and, at the same time, the presence, possibly the type and risk of a defect. For example, the point of maximum amplitude of this curve, which corresponds to the edge of the protrusion furthest from the vertical axis Z of the container, can be considered. This maximum amplitude corresponds to the height of the peak of the deviation curve CEc, which gives an accurate estimate of the radial length of the knockout or flange and constitutes a very good physical measure of the risk of a defect. This amplitude can be compared with a threshold value, and if this measurement value exceeds this threshold, a deviation (or defect) is detected.

[0092] The threshold is adjustable to parameterize the stringency of the inspection; for example, the operator can decide that a slightly marked knockout is acceptable.

[0093] If the measurement value that allows the deviation to be detected is the area of ​​the surface obtained by subtraction of the contour of the end of the finish from the reference contour of the end of the finish, the measurement value is, for example, the integral of the deviation curve CEc. If the measurement value is a gradient, it can be the maximum gradient of the deviation curve CEc or the deviation between two consecutive gradients of the deviation curve CEc on either side of a peak or a threshold crossing.

[0094] From the above description, it can be seen that the method according to the present invention is capable of detecting at least one horizontal mold defect selected from the group consisting of knockout JHK, flange JHF, and overhang / overmatch JHO. seam It is clear that this involves detecting JH defects.

[0095] According to an advantageous variant of the embodiment, the method according to the invention makes it possible to distinguish between the detected defects, namely, knockout JHK, flange JHF, and overhang / overmatch JHO. For this purpose, at least one detection criterion is defined for each defect, namely, knockout, flange, and overhang / overmatch, and at least one of these detection criterion is used to identify the defect. This makes it possible to identify the type of defect, which may improve the accuracy of the container inspection.

[0096] According to one exemplary embodiment, the shape of the deviation detected for several elevations in the inspection zone is selected as the detection criterion for at least one image. For example, the shape of the deviation can be described in various ways, for example, using measurements of slope or amplitude at several consecutive elevations. This shape can be observed in the deviation curve CEc. In FIG. 9, the deviation curve CEc has a peak that characterizes a knockout or flange, whereas if the defect is an overhang / overmatch as in FIG. 5A, the deviation curve CEc is as shown in FIG. 10. It becomes a staircase shape that can be seen.

[0097] According to another exemplary embodiment, a deviation observation angle range corresponding to the number of consecutive images is selected as the detection criterion, and the deviation is detected there, i.e., at the container rotation angle during which the deviation is detected in the images. As can be clearly seen in FIGS. 3B, 4B, and 5B, the deviation observation angle range for flange or overhang / overmatch type defects is very close to the circumferential angle range of the defect viewed from above. For a knockout, the observation angle range is greater than its circumferential angle range. For example, if the rotation increment between two images is 4° and the knockout has a very small circumferential angle range of 2°, and if it is long enough, it can be seen in several consecutive images separated in time by one or more rotation increments, respectively, resulting in an observation angle range of 8° or 12°, respectively.

[0098] Of course, these criteria can be implemented independently of each other or in combination, one after the other, and one of these criteria can be applied before the other, and vice versa.

[0099] According to one example of implementation, the shapes of the detected deviations for several elevations in the inspection zone are analyzed to identify overhang / overmatch JHO defects versus knockout JHK and flange JHF defects. Indeed, as can be clearly seen from Figure 5A, overhang / overmatch JHO defects are identified in particular by their staircase or rampart shape versus knockout (Figure 3A) and flange (Figure 4A) defects, which are characterized by sharp protrusions.

[0100] According to another example of implementation, the observation angle range of the deviation is analyzed and a knockout JHK defect is identified relative to a flange JHF defect, taking into account that a knockout defect is detected if the observation angle range of the deviation is below a maximum value, for example set at 30°. In other words, a knockout JHK defect is detected if the number of consecutive images in which the deviation is detected corresponds to an observation angle range below a maximum value, for example set at 30°.

[0101] According to another example of implementation, the observation angle range of the deviation is analyzed and a knockout JHK defect is distinguished from a flange JHF defect, taking into account that a flange defect is detected if the observation angle range of the deviation exceeds a minimum value, e.g., 30°. In other words, a flange JHF defect is detected if the number of consecutive images in which the deviation is detected corresponds to an observation angle range that exceeds a minimum value, e.g., 30°.

[0102] According to the present invention, discrimination criteria such as thresholds applied to the measurements to detect deviations, those applied to the comparison of the contour shape and / or those applied as minimum or maximum values ​​for the observation angle range are preferably stored in the analysis processing unit 5 and adjustable by the operator depending on the inspected container and the required quality.

Claims

1. 1. A method for detecting defects in horizontal mold seams (JH) for finishes (B) of a plurality of glass containers (R), each having a vertical axis (Z), the method for detecting the defects for each container comprising the steps of: placing the container (R) between a light source (3) and an image capture camera (4), wherein the observation optical axis of the image capture camera is substantially perpendicular to an axis parallel to the vertical axis (Z) of the container, and the field of view of the image capture camera includes at least the left edge of the finish or the right edge of the finish, including at least a portion of the horizontal mold seam; Ensuring rotation of said container (R) along said vertical axis (Z) through at least one revolution; acquiring an image by said camera (4) at each rotational increment of said container such that the number of images per rotation is greater than 36; and analyzing the captured images for each container; Including, The image is a finish inspection zone (ZI) spanning the height of the finish including at least a portion of the horizontal mold seam as defined in the image; The contour (P) of the end of the finish is detected in the image inspection zone. comparing the contours (P) of the edge of the finish of the image with a reference contour (Pf) of the edge of the finish to detect deviations between these contours of the edge of the finish and the reference contour of the edge of the finish; and a defect in the horizontal mold seam (JH) of a container is detected when at least one image of the container has a deviation; It has been captured, The method includes distinguishing the detected defects from among the knockout (JHK) and the flange (JHF) based on at least one detection criterion, wherein the at least one detection criterion is an observation angle range of the deviation corresponding to a number of consecutive images in which the deviation is detected. method.

2. 2. The method of claim 1, wherein the camera (4) is positioned so that the field of view includes at least a portion of the horizontal mold seam (JH) and a fiducial whose elevational position relative to the horizontal mold seam is known and defines the finish inspection zone (ZI) in the image, the fiducial whose elevational position relative to the horizontal mold seam is identified, and the finish inspection zone (ZI) is positioned relative to the fiducial such that the finish inspection zone extends over a height that includes the horizontal mold seam.

3. 3. The method of claim 2, wherein all or part of the finished surface (S) or of the unfinished surface (CB) is identified in the image as a reference.

4. 4. The method according to any one of claims 1 to 3, wherein the image capture camera (4) has a horizontal width of the field of view of 5 mm to 130 mm and a height of the field of view of 3 mm to 20 mm.

5. 5. The method according to any one of claims 1 to 4, wherein the image capture camera (4) is positioned so that its observation optical axis is substantially tangent to the left or right edge of the finish.

6. 6. The method according to any one of claims 1 to 5, wherein there is present a light source (3) having vertical and horizontal illumination dimensions of 100% to 200%, preferably 100% to 120%, of the field of view dimensions of the camera (4) multiplied by the distance between the light source and the lens (4a) of the camera and divided by the distance between the axis of the container and the lens of the camera.

7. 7. A method according to any one of claims 1 to 6, wherein the contour (P) of the end of the finish in the image is compared with the reference contour (Pf) of the end of the finish by using as reference contour (Pf) at least one contour obtained for at least one reference container that is considered to be suitable, and deviations are detected.

8. 7. A method according to any one of claims 1 to 6, wherein the contour (P) of the end of the finish in the image is compared with the reference contour (Pf) of the end of the finish by using as the reference contour (Pf) an average contour calculated over several rotational increment values ​​of at least one reference container that is considered to be compatible, and deviations are detected.

9. 9. A method according to any one of claims 1 to 8, wherein the contour (P) of the end of the finish in the image is compared with the reference contour (Pf) of the end of the finish by using the contour of the end of the finish to which a low pass filter has been applied as a reference contour (Pf), and deviations are detected.

10. 7. A method according to any one of claims 1 to 6, wherein the contour (P) of the end of the finish in the image is compared with a reference contour (Pf) of the end of the finish by applying a high pass filter to the contour of the end of the finish, and deviations are detected.

11. 11. A method according to any one of claims 1 to 10, wherein for each image, the contour (P) of the end of the finish and the reference contour (Pf) of the end of the finish are compared by comparing at least one of the area, amplitude and / or slope measurements with a threshold value for at least some heights in the inspection zone, and a deviation is detected if at least one of these measurements exceeds this threshold value.

12. 12. The method according to any one of claims 1 to 11, wherein for at least one image the shape of the deviations detected for several heights in the examination zone is selected as detection criterion.

13. 13. The method of claim 1, wherein the shapes of the deviations detected for several elevations in the inspection zone are analyzed to identify overhang / overmatch (JHO) defects from the knockout (JHK) and flange (JHF) defects.

14. 14. The method according to claim 1, wherein the observation angle range of the deviation is analyzed to distinguish the knockout (JHK) defect from the flange (JHF) defect by considering a knockout defect to be detected if the observation angle range of the deviation is below a certain maximum value.

15. 15. The method according to claim 1, wherein the observation angle range of the deviation is analyzed to distinguish the knockout (JHK) defect from the flange (JHF) defect by considering a flange defect to be detected if the observation angle range of the deviation exceeds a certain minimum value.

16. 16. An inspection device configured to perform the method of any one of claims 1 to 15 for detecting defects in the horizontal mould seam (JH) of the finish of the glass container.

Images