Inspection of tubular products

Abstract

Apparatus for inspecting a plurality of tubular products is disclosed The tubular products are in a container which may have a predetermined packing configuration. The apparatus comprises a first camera (30) arranged to capture a first image of an inside of a tubular product (10) with a first part of the tubular product substantially in focus, and a second camera (32) arranged to capture a second image of the inside of the tubular product with a second part of the tubular product, different from the first part, substantially in focus. Means (50) are provided for analysing the first and second images to determine whether there is a defect in the tubular product. This may allow inspection of the tubular products to take place at speeds which are consistent with the manufacturing process.

Description

[0001] INSPECTION OF TUBULAR PRODUCTS

[0002] The present invention relates to techniques for the inspection of tubular products, and in particular but not exclusively collapsible tubes such as those used in the packaging of consumer products and pharmaceuticals.

[0003] A collapsible tube is a package which can be used as a means of containing and dispensing products such as toothpaste, cosmetics, paints, adhesives, foodstuffs, healthcare products, pharmaceuticals, etc. Typically, but not exclusively, the tube comprises a hollow cylindrical sleeve with a round or oval profile. The sleeve is closed at one end by a shoulder with an orifice which allows the product to be dispensed. The other end of the tube is left open to allow the tube to be filled.

[0004] The process of manufacture of the tube is normally separated from the process of filling. The tubes are made and then packed into a box of known dimensions with a particular packing pattern. The box of open tubes is transported to the filling operation where the tubes are unpacked and loaded into a filling machine. Once filled, the open ends of the tubes are sealed.

[0005] One challenge in the tube production process is ensuring that the tubes are free of manufacturing defects, in the correct packing configuration, complete in all respects and free of contamination. This is particularly critical for the inside of a tube where the filling is in contact with the tube walls.

[0006] Existing quality measures in the tube production process are in the main restricted to good practices within the manufacturing environment and sampling for inspection. There are batch tests for systematic construction faults such as poor shoulder adhesion which is a consequence of incorrect machine settings. However, these are slow and destructive in nature. Visual inspection may be performed for contamination and other defects. However, this process may be compromised by poor lighting, fatigue, subjectivity and human error.

[0007] It has been proposed to use a camera to inspect tubes in some manufacturing environments. However, a number of problems have been encountered. Firstly, the implementations have been slow and do not meet some manufacturing requirements. This is due in part to the variation in product length which may prevent the camera from capturing a sufficiently high quality image of the inside length of the tube. Furthermore, a small aperture may be needed in the camera lens to obtain a sufficiently large depth of field. This may require an extended exposure time for the camera to obtain good quality images, further slowing the inspection process beyond what is acceptable. Since the tube making process is usually continuous, inspection should not impact that process by being slower than the rate of manufacture. Another barrier to adoption is the classification of defects such that rejection is made on a pragmatic basis, taking into account the significance of the defect or contamination.

[0008] It would therefore be desirable to provide a more complete, consistent and effective technique for the non-destructive inspection of tubes. In particular, it would be desirable to provide a system which can identify both random and process faults without slowing the inspection process.

[0009] According to one aspect of the present invention there is provided apparatus for inspecting a tubular product, the apparatus comprising: a first camera arranged to capture a first image of an inside of the tubular product with a first part of the tubular product substantially in focus; a second camera arranged to capture a second image of the inside of the tubular product with a second part of the tubular product, different from the first part, substantially in focus; and means for analysing the first and second images to determine whether there is a defect in the tubular product.

[0010] The present invention may provide the advantage that, by capturing a first image with a first part of the tubular product substantially in focus and a second image with a second part of the tubular product substantially in focus, it may be possible to identify defects in different parts of the tubular product and / or different types of defect, such as random and process faults, while maintaining a desired inspection speed. This may allow inspection of tubular products to take place at speeds which are consistent with the manufacturing process. It will be appreciated that the areas of interest (parts of the tubular product) do not have to be exactly in focus, as long as they are sufficiently in focus to allow image analysis to take place. Furthermore, there may be some overlap between the areas which are in focus in the two images, or in some cases an area which is in focus in one image may lie wholly within an area which is in focus in the other image (for example, where one image has a large or infinite depth of field). Thus, an area which is in focus in one image may be contained within, or overlap with, or be contiguous with, or touch, or be adjacent to, or be separate from an area which is in focus in the other image.

[0011] The tubular product may comprise one end which is at least partially open and another end which is at least partially closed. For example, one end may be (at least partially) open to allow the tubular product to be filled. The other end may comprise a shoulder, an orifice (such as a nozzle) and / or a cap which may at least partially close that end of the tubular product. A sleeve may be provided between the open and closed ends. The sleeve may form a chamber for containing a product with which the tubular product is to be filled. The sleeve may be in the form of a hollow cylinder (for example, a circular or elliptical cylinder) or it may be frustoconical or any other appropriate shape. The sleeve may comprise an inner wall. The tubular product may have a length (along its longitudinal axis) which is greater than its diameter. For example, the tubular product may have a length to diameter ratio of at least 2, 3, 4 or 5, or any other suitable value. The first camera and the second camera may be arranged to view the inside of the tube through the (at least partially) open end of the tubular product.

[0012] Preferably the first camera is focused on the first part of the tubular product and the second camera is focused on the second part of the tubular product. For example, the first camera may have a first focus point or focus area, and the second camera may have a second focus point or focus area, different from the first. This may be achieved by arranging the first and second cameras to have different focal lengths. Thus, the first camera may have a first focal length, and the second camera may have a second focal length different from the first focal length. Alternatively, or in addition, the first and second cameras may be arranged at different distances from the tubular product. This may allow the cameras to be focused on different parts of the tubular product.

[0013] The first camera may be focused on an at least partially closed end of the tubular product (within a margin of error). For example, the first camera may be focused on a shoulder, nozzle and / or cap of the tubular product. Thus, the first image may be an image in which the at least partially closed end of the tubular product is substantially in focus. This may allow the image of the (at least partially) closed end of the tube to be of sufficiently high quality to allow defects, such as faults or contamination, in that end of the tube to be identified.

[0014] The second camera may be focused on at least part of an inner wall of the tubular product. For example, where the tubular product has a sleeve, the second camera may be focused on at least part of the inner wall of the sleeve. Thus, the second image may be an image in which at least part of the inner wall of the tubular product is in focus. This can allow the image of the inner wall to be of sufficiently high quality to allow defects, such as such as faults or contamination, in the inner wall / sleeve to be identified.

[0015] The second camera may have a different depth of field from the first camera. For example, the second camera may have a depth of field such that at least a majority of an inner wall of the tubular product is substantially in focus. In this case, at least 50%, 60%, 70%, 80% or 90% of the inner wall of the tubular product may be substantially in focus, although other values are also possible. This may allow defects (such as faults or contamination) along substantially the whole of the inner wall to be identified. On the other hand, the first camera may have a depth of field such that at least a majority of the (at least partially) closed end of the tubular product (for example, a shoulder, nozzle and / or cap) is substantially in focus. This can allow defects in the closed end of the tubular product to be identified reliably.

[0016] In general, the sleeve of the tubular product will be longer (in the direction of its longitudinal axis) than the closed end (shoulder, nozzle and / or cap). Thus, the second camera may have a depth of field which is greater than that of the first camera. For example, the second camera may have a depth of field which is at least 2, 3, 4 or 5 times greater than the first camera, although other values are also possible. In one embodiment, the second camera is a pin-hole type camera with a large or infinite depth of field, although other types of camera could be used instead.

[0017] The apparatus may further comprise means for providing relative movement between the cameras and the tubular product. For example, the first and second cameras themselves may be moved, or the tubular product may be moved, or both. The movement may be in two (or more) dimensions. This may allow the tubular product to be brought into a field of view of one or both of the cameras.

[0018] The apparatus may further comprise control means for controlling the relative movement between the cameras and the tubular product. The control means may be arranged to control the relative movement such that the tubular product is brought into a field of view of one or both of the cameras.

[0019] In some cases, the cameras may have different fields of view. For example, the cameras may be adjacent to each other, and may have adjacent fields of view. The cameras may for example be co-axial with each other, that is, they may have lines of sight which are substantially parallel, or they may have lines of sight that are at an angle to each other and / or to a longitudinal axis of the tubular product. Where the cameras have different fields of view, the control means may control relative movement between the cameras and the tubular product such that the tubular product is in a field of view of the first camera and a field of view of the second camera sequentially in time (in either order). Alternatively, in some cases it may be possible for the tubular product to be in the field of view of both cameras simultaneously.

[0020] In some cases, individual tubular products may be inspected. For example, it would be possible for individual tubular products to be sampled (for example, from a production line) and inspected. However, in preferred embodiments, the apparatus is arranged to inspect a plurality of tubular products as a group. Thus, the apparatus may be arranged to inspect a plurality of tubular products. In this case, the means for providing relative movement may be arranged to provide relative movement between the cameras and the plurality of tubular products. The control means may be arranged to control the relative movement to bring the tubular products into a field of view of the cameras. For example, each camera may have a different field of view, and the cameras may be arranged such that, at any one time, a tubular product may be in the field of view of one camera but not the other. The control means may be arranged to control the relative movement such that each of the plurality of tubular products is brought into a field of view of each of the cameras sequentially.

[0021] In one embodiment, the means for providing relative movement is arranged to provide continuous movement between the cameras and the plurality of tubular products. In this case, the first and second cameras may be arranged to capture the first and second images during the continuous movement. For example, the images may be captured and / or a light source may be triggered as a tubular product is brought into a field of view of a camera. This may facilitate high speed inspection of the tubular products.

[0022] The plurality of tubular products may be provided in a container, such as a box, crate or tray. The container may be used to transport the tubular products, for example, to another facility or another part of the manufacturing process. Thus, the apparatus may be arranged to inspect a plurality of tubular products in a container. In this case, means may be provided for bringing the tubular products in the container into the fields of view of the first camera and the second camera.

[0023] The tubular products may be packed in the container with a predetermined packing configuration. In this case, the control means may have knowledge of the predetermined packing configuration. For example, the packing configuration may be entered by an operator, or may be determined automatically (for example, using a camera or by scanning a barcode), or may be provided by another entity, or may be fixed. Knowledge of the packing configuration may then be used to move the cameras to suitable positions for imaging the tubular products (such as over the centres of the tubular products). Thus, the control means may be arranged to control relative movement between the cameras and the tubular products in accordance with the predetermined packing configuration. For example, in one embodiment the control means may cause the first and second cameras to scan sequentially over the plurality of tubular products.

[0024] The apparatus may further comprise means for determining the packing configuration. For example, in one embodiment, the apparatus may comprise an auxiliary camera which is arranged to take an image of all of part of the plurality of tubular products (or at least one of the first and second cameras may be used for this purpose). In this case, the image may be analysed to identify the packing configuration and / or type of tube.

[0025] Where the tubular products are provided in a container, the container may be conveyed into the apparatus so that it is in a suitable position for inspection of the tubular products. Thus, the apparatus may further comprise means for conveying a container with a plurality of tubular products into the inspection apparatus. The apparatus may further comprise a datum stop which may be used to locate the container relative to the apparatus. The datum stop may be used as a reference (datum) for subsequent movement of the cameras relative to the tubular products. Thus, the control means may be arranged to move the cameras relative to the tubular products using the datum stop as a reference. This may help to ensure that the cameras are in the correct positions relative to the tubular products and / or triggered at the correct times.

[0026] The tubular products may have a predetermined diameter, which may be fixed or variable. In this case, the control means may have knowledge of the predetermined diameter. For example, the diameter may be entered by an operator, or it may be determined automatically (for example, by analysing an image of the product, which image may taken by the first camera or the second camera or by another camera), or the diameter may be provided by another entity, or it may be fixed. Knowledge of the predetermined diameter may be used to ensure that the cameras are suitably positioned relative to the tubular products when capturing images. Thus, the control means may be arranged to move the cameras relative to the tubular products in accordance with the predetermined diameter of the tubular products. For example, in some embodiments, the cameras may be positioned on or around a longitudinal axis of the tubular products when capturing images. Knowledge of the packing configuration and / or diameter of the tubular products may be used to trigger the cameras (and / or light sources) when it is predicted that a tubular product is in a field of view of a camera. Thus, the control means may be arranged to trigger the first camera to take the first image when it is predicted that the tubular product is in a field of view of the first camera, and to trigger the second camera to take the second image when it is predicted that the tubular product is in a field of view of the second camera.

[0027] In one embodiment, one or more of the cameras is arranged to capture a burst of images when it is triggered. The burst of images may comprise at least 2, 3, 4 or 5 images (for example, between 3 and 5 inclusive) although it may be more. Thus, at least one of the first and second cameras may be arranged to capture a plurality of images of the tubular product when it is triggered (for example, when it is predicted that the tubular product is in a field of view of the camera). In this case, the analysing means may be arranged to select at least one of the plurality of images for analysis. For example, the analysing means may be arranged to select an image in which the whole of the tubular product is visible and / or in which the tubular product is centrally located. This may help to compensate for any offset between the expected and actual position of the tubular product which may occur, for example, if there is continuous movement between the camera and the tubular product while the images are taken. However, if desired, other factors such as image quality may be used to select the image as well or instead.

[0028] In other embodiments, the camera is arranged to take images continuously (for example, in the form of a video). In this case, the analysing means may be arranged to select one or more images from a continuous stream of images for analysis.

[0029] The apparatus may further comprise at least one light source arranged to at least partially illuminate a field of view of the first camera and / or the second camera. This may help to ensure that the inside of the tube is sufficiently illuminated to allow images of sufficient quality to be captured. For example, the apparatus may comprise a first light source arranged to at least partially illuminate a field of view of the first camera and / or a second light source arranged to at least partially illuminate a field of view of the second camera (although it would also be possible to use the same light source for both cameras). The first and / or second light sources may be, for example, ring lights, and may be coaxial with the respective cameras, although other arrangements such as side lights are also possible.

[0030] The first and / or second light sources may be pulsed light sources. In this case, the first light source may be triggered when an image is taken by the first camera and / or the second light source may be triggered when an image is taken by the second camera. For example, the control means may be arranged to trigger the first and / or second light source when it is predicted that the corresponding camera has the tubular product in its field of view.

[0031] In some embodiments, the first light source may have a first intensity, and the second light source may have a second intensity different from the first intensity. For example, the second light source may have a higher intensity than the first light source. It has been found that reflections from the closed end of the tubular product may be more prominent than reflections from the sidewall. Thus, by using lower intensity light to capture the first image, reflections from the closed end of the tube may be reduced. This may facilitate processing of the first image to identify defects in the closed end. On the other hand, by using higher intensity light to capture the second image, it may be possible to obtain a sufficiently high- quality image of the sidewall to allow defects in the side wall to be identified. For example, using a higher intensity light source for the second camera may allow the second camera to have a greater depth of field than the first camera (for example, by reducing the size of the aperture) and / or avoid extended exposure times.

[0032] The defect in the tubular product may be, for example, a manufacturing fault and / or contamination. Furthermore, where a plurality of tubular products are provided in a container, the defect may be a packing fault such as a missing or misplaced tubular product. Thus, the analysing means may be arranged to analyse the first and second images to determine whether there is a manufacturing fault and / or a packing fault and / or contamination in the tubular product. This may allow defects which may lead to rejection of the tubular product to be identified, and may allow the tubular product to be classified, for example, as accept or reject, or as requiring further inspection.

[0033] The analysing means may be arranged to analyse the first image to determine whether there is a defect in the closed end of the tubular product. The defect may be a defect which occurs during manufacture, transport or packing. For example, the defect may be one or more of: a faulty joint between the shoulder and the sleeve; a missing cap; a missing seal; a fault in the shoulder, nozzle, cap and / or seal; contamination; or any other fault. In general, such faults will be to tend to be process faults. Such faults may be identified with a reasonable degree of accuracy by analysing an image focused on the closed end of the tubular product.

[0034] The analysing means may be arranged to analyse the second image to determine whether there is a defect in an inner wall of the tubular product. For example, the defect may be one or more of: a fault in a sidewall seam; contamination of the inner wall; and any other defect. Such defects may be identified with a reasonable degree of accuracy by analysing an image of the inner wall captured with a relatively large depth of field.

[0035] The analysing means may be arranged to analyse the first image and / or the second image to determine a physical property, such as ovality and / or diameter, of the tubular product. This may be useful in determining whether the tubular product should be classified as accept or reject, or requiring further inspection.

[0036] The analysing means may be arranged to annotate the first image and / or the second image to indicate potential defects. For example, the analysing means may be arranged to analyse the first image and / or the second image to identify an unexpected feature (i.e. a feature which would not be expected in an acceptable product), and to highlight the part of the image with the unexpected feature. This may facilitate the identification of faults by an operator. For example, the annotated first image and / or second image may be displayed on a display which may be viewed by the operator. The apparatus may further comprise means for classifying the tubular product as accept or reject in dependence on an output of the analysing means. For example, the classification means may classify the tubular product as reject if it is determined that there is a manufacturing fault and / or if it is determined that there is contamination and / or if a physical property of the tubular product, such as ovality or diameter, is outside of a predetermined range and / or if it is determined that there is a packing fault such as a missing or misplaced tubular product. In some cases, individual faults may not in themselves cause the tubular product to be classified as reject, but an accumulation of such faults or other faults may do so. On the other hand, the classification means may classify the tubular product as accept, for example, if it is determined that there is no significant manufacturing fault, substantially no contamination, if the physical property of the tubular product is inside a predetermined range and / or if there is no packing fault. It will be appreciated that other classifications, such as requiring further inspection, are also possible.

[0037] The apparatus may further comprise means for producing a reject signal if the tubular product is classified as reject. The reject signal may for example cause an audible or visual alarm to sound, or it may be transmitted to another apparatus such as a machine for making the tubular products, or it may be conveyed to the operator in any other way such as via a user interface. This can allow the appropriate action to be taken in response to the identification of a defect.

[0038] In some embodiments, three or more cameras may be used to capture images of the tubular product. For example, the apparatus may comprise, three, four, five or more cameras. This may allow images to be captured with different parts of the tubular product in view and / or in focus.

[0039] In some embodiments, at least one of the cameras has a line of sight which is at an angle to a longitudinal axis of the tubular product. For example, a camera which is arranged to capture an image of an inner wall of the tubular product may have a line of sight at an angle to the longitudinal axis. This may be achieved, for example, by tilting the camera and / or using the Scheimpflug principle (for example, tilting the camera lens and / or using a Scheimpflug adaptor). This may allow a more detailed image of the inner wall to be captured. In one embodiment, a plurality of cameras (for example, two, three, four or more) are provided with a line of sight at an angle to a longitudinal axis of the tubular product. In this case, the directions in which the lines of sight are angled may be spaced circumferentially around the tubular product. For example, the directions in which the lines of sight are angled may be spaced evenly around the tubular product in a circumferential direction. This may allow more detailed images of different parts of the inner wall to be captured. The amount by which the lines are sight are angled may be the same or different for each camera.

[0040] In alternative embodiments, a single camera may be used to capture both images, for example with the focus and / or depth of field adjusted between the capturing of the images.

[0041] In any of the arrangements described above, the tubular product may be a packaging product, and may be of a type which is arranged to be filled with a dispensable product. The dispensable product may be a viscous liquid, and may be dispensable through an orifice in a closed end of the tubular product. The dispensable product may be, for example, a pharmaceutical, cosmetic, food, consumer and / or industrial product or any other suitable product. In one embodiment, the tubular product is a collapsible tube which is manufactured with one end open prior to filling. However, the invention is applicable to any packaging or tubular product where there is a need to look at the inside of the package walls and at the inside end of the package e.g. ampules or vials.

[0042] Corresponding methods may also be provided. Thus, according to another aspect of the invention there is provided a method of inspecting a tubular product, the method comprising: capturing a first image of an inside of the tubular product with a first part of the tubular product substantially in focus; capturing a second image of the inside of the tubular product with a second part of the tubular product, different from the first part, substantially in focus; and analysing the first and second images to determine whether there is a defect in the tubular product. The first image may be captured with a first camera and the second image may be captured with a second camera, although it would also be possible to use the same camera to capture both images.

[0043] According to a further aspect of the invention there is provided apparatus for inspecting a plurality of tubular products, the apparatus comprising: at least one camera arranged to capture images of the tubular products; means for providing relative movement between the camera and the tubular products; means for controlling the relative movement to bring the tubular products into a field of view of the camera; and means for analysing the captured images to determine whether there is a defect in the tubular product, wherein the plurality of tubular products are provided in a container with a predetermined packing configuration, and the control means is arranged to control the relative movement in accordance with the predetermined packing configuration.

[0044] In one embodiment, an auxiliary camera is provided which is arranged to take an image of all of part of the plurality of tubular products. In this case, the image may be analysed to identify the packing configuration. Alternatively or in addition, the predetermined packing configuration may be entered by an operator, or may be determined automatically, for example, by scanning a barcode or other feature, or may be provided by another entity, or may be fixed.

[0045] The control means may be arranged to trigger the camera to take an image (and / or trigger a light source) when it is predicted that the tubular product is in a field of view of the camera,

[0046] Features of one aspect of the invention may be used in combination with any other aspect. Any of the apparatus features may be provided as method features and vice versa. Preferred features of the present invention will now be described, purely by way of example, with reference to the accompanying drawings, in which:

[0047] Figure 1 shows schematically a sectional side view of a tube;

[0048] Figure 2 shows schematically an end view of a tube;

[0049] Figures 3(A) to 3(C) show schematically different packing patterns for tubes packed in a box;

[0050] Figure 4 shows a side view of a tube inspection system in an embodiment of the invention;

[0051] Figure 5 shows in more detail parts of the tube inspection system;

[0052] Figure 6 shows a top view of the tube inspection system;

[0053] Figure 7 shows parts of a tube inspection system in an embodiment of the invention;

[0054] Figure 8 shows parts of the parts of an image processing unit in more detail;

[0055] Figures 9 and 10 show typical images from a first camera and a second camera respectively;

[0056] Figure 11 parts of a tube inspection system in another embodiment;

[0057] Figure 12 shows steps taken by a control unit to scan the cameras over the tubes in one embodiment;

[0058] Figure 13 shows steps taken by a first image processing module in one embodiment;

[0059] Figure 14 shows steps taken by a second image processing module in one embodiment; and

[0060] Figure 15 shows steps carried out by a classification unit in one embodiment.

[0061] Collapsible tubes are used widely as a means of containing and dispensing products such as toothpaste, cosmetics, paints, adhesives, foodstuffs, healthcare products and pharmaceuticals. Typically, the tube comprises a cylindrical, hollow body with a round or oval profile. At one end of the tube body there is usually an orifice, which can be closed by different caps and closures. The orifice can be shaped in different ways, and is typically in the form of a nozzle in various styles and lengths. The other end is sealed for example by welding or by folding. Collapsible tubes may be made from various materials including metals (such as aluminium), plastics (such as polyethylene), plastic / aluminium laminates (such as aluminium barrier laminate (ABL) or plastic barrier laminate (PBL)), paperboard, or any other suitable materials. The tubes may be of variable diameter, typically between approximately 10mm to >50mm, and of variable length, typically 25mm to >300mm. The tubes may be manufactured in various ways, such as by means of an extrusion process, or by forming a sleeve and then welding on a plastic shoulder, or by injection moulding, or any other suitable process. The speed of manufacture of tubes varies both by machine and materials of the tube but can vary from tens per minute up to 600 tubes per minute for the fastest laminate tube manufacturing machines currently available.

[0062] Prior to filling, the tubes are usually fully open at one end, and closed by some form of shoulder and / or cap at the other end. The manufacture process may involve placing a film over the cap end of the tube or fitting a cap of various forms.

[0063] Figure 1 shows schematically a sectional side view of a tube in one example. Referring to Figure 1 , the tube 10 comprises sleeve 12, shoulder 14, nozzle 15, optional tamper evident seal 16 and cap 18. The sleeve 12 may be formed as part of an extrusion or moulding process, or it may be formed by rolling a sheet of material and forming a seam 20, or in any other way. The shoulder 14 may be an integral part of the tube and may be formed for example as part of an extrusion process. Alternatively, the shoulder may be a separate component and may be welded to the sleeve, or attached in some other way. The shoulder 14 and nozzle 15 are provided at one end of the sleeve 12. The other end of the sleeve is left open to allow the tube to be filled. The tube has a longitudinal axis as indicated by the dashed line.

[0064] Figure 2 shows schematically an end view of the tube of Figure 1 . In Figure 2, the inside of the tube is viewed through its open end. The rim 21 of the tube and the inside wall 22 of the sleeve 12 are visible, as well as the shoulder 14 and the nozzle 15. The formed tube manufacture in most cases is a subcontracted process from the filling operation. In this case the tubes are made and then packed, often automatically, into a box of known dimensions with a particular packing pattern. The box of open tubes is then transported to the filling operation where the tubes are unpacked, normally robotically, and loaded into the filling machine.

[0065] Figures 3(A) to 3(C) show schematically some different packing patterns for tubes packed in a box. Each box 24 is packed with a plurality of tubes 10. The tubes are viewed from their open ends, looking downwards into the insides of the tubes. In Figure 3(A) the tubes are packed in an ordinary or square pattern, while in Figures 3(B) and 3(C) the tubes are packed in a hexagonal pattern. The relative box dimensions required for the different patterns and the number of tubes (N) that can be packed in these configurations for these similar box sizes are shown.

[0066] During the manufacturing, packing and shipping process, various faults or defects may occur. For example, the tube may include a tamper evident seal over the end of the tube proper between the tube and cap. This provides a hermetic seal between the future contents of the tube and the outside world. This seal can be missing or not fully covering the entrance to the tube. A missing cap would also be a defect. Where the tubes include a seam, the overlap and thickness of this seam is important to the integrity of the tube. Any material that is trapped by the heat forming of the seam (such as by ultrasonic welding), will compromise the seam strength. It is desirable to reject tubes with inconsistent seam welds or welds with trapped material. Ovality of the tube is another inspection point.

[0067] Furthermore, debris adhering to the inside of the tube or at the shoulder would be defect. In addition, one or more tubes may be missing or misplaced.

[0068] One challenge for the tube production process is the demands of the filling process for “perfect” tubes, free of manufacturing defects, in the correct packing configuration, complete in all respects and free of contamination. This is particularly critical for the inside of a tube where the filling is in contact with the tube walls. The latter is increasingly becoming a critical inspection / rejection criterium for incoming tubes in not only foodstuff and pharmaceutical applications but also for cosmetic preparations. Contamination could be in the form of dust, card, or other particulates on the inside of the tube that have been deposited there from the manufacturing or transportation environment. This is exacerbated by the tendency of plastic and laminate tubes to accumulate electrostatic charge which attracts airborne particles. As contamination is likely to be a random fault occurrence rather than a process fault (such as running out of caps) the only means to eliminate this fault is through inspection of all tubes (100% inspection) and rejection of non-conforming tubes.

[0069] Existing quality measures are in the main restricted to good practices within the manufacturing environment and sampling for inspection. There are batch tests for systematic construction faults such as poor shoulder adhesion which is a consequence of incorrect machine settings. These are slow and destructive in nature. Currently, contamination inspection is performed visually by an operator who is expected to scan a box of tubes and reject those that are not suitable for shipping. However, the effectiveness of human tube inspection may be compromised by poor lighting, fatigue and subjectivity. The consequences of shipping faulty tubes can be quite severe as some filling companies will reject a batch of many thousands of tubes if a single tube is found to be faulty by their goods inwards inspectors.

[0070] A more complete, consistent and effective method of 100% tube inspection for contamination and construction defects by non-destructive means would be a significant advantage to the tube manufacture industry and the goods inspection process at the receiving / filling company.

[0071] Embodiments of the invention provide a system for the inspection of collapsible tubes / plastic tubes or laminate tubes that are used in the packaging of consumer products and pharmaceuticals. The inspection system is devised to quickly scan and inspect a box of empty (unfilled) tubes before being either sent to the next stage in the manufacture process or being loaded on a filling machine. The system can operate at high speeds and principally inspects the inside of the tube. Inspection is made on the basis of high resolution images of high quality that enable the discrimination of small particles from the background of the tube.

[0072] Embodiments of the invention deploy two cameras, each with a synchronised high intensity pulsed light and a gantry that allows movement of the cameras and lights. The travel of the cameras with respect to the tube centres is such that the cameras move true to the tube centres.

[0073] Figure 4 shows a side view of a tube inspection system in an embodiment of the invention. Referring to Figure 4, the system comprises a gantry 26 that has a motion track 28 on which are mounted two cameras 30, 32. The cameras 30, 32 are digital cameras and include light sensors such as Charge-Coupled Devices (CCDs) or Complementary Metal-Oxide-Semiconductor (CMOS) devices which convert incoming light into electrical signals to form digital images. Two light source 34, 35 are provided, each of which is arranged to illuminate a field of view of one of the cameras 30, 32. Each light source 34, 35 is a ring light which is coaxial with the respective camera 30, 32. The light sources 34, 35 are pulsed light sources, which emit a flash of light in response to a trigger. An infeed conveyer 36 is provided to feed boxes 24 of tubes into the gantry 26. A datum stop 38 is provided to locate the boxes 24. The motion track 28 can move the cameras 30, 32 in an x-y motion over the tubes 10 in their box 24 using servomotor driven linear actuators 40. The cameras 30, 32 have a line of sight in the z-direction, downwards into the open tubes.

[0074] In the arrangement of Figure 4, the first camera 30 is fitted with a fixed focal length lens, and is focussed so that an image can be formed of the region of the tube shoulder and cap (the bottom of the tube). The second camera 32 is alongside the first camera 30 and faces in the same direction. The second camera 32 has a smaller aperture than the first camera, and may be of a “pinhole” type or similar. This gives a large depth of field but requires larger exposure times or more intense lighting conditions to form an image with high definition. The second light source 35 produces relatively high intensity lighting to allow the second camera 32 to fully utilise its large depth of field without extending exposure times.

[0075] In operation, after the tubes are made and packed into a box, the box 24 is fed automatically into the gantry 26 using the infeed conveyor 36. Movement is mainly in the x-direction in Figure 4, although there may also be some movement in the y- and / or z-direction to accommodate differently sized boxes and / or tubes. The box 24 is located using the datum stop 38. The cameras 30, 32 are then scanned over the box of tubes by means of the servomotor driven linear actuators 40. A series of images is acquired of the shoulder / bottom of the tubes and the inside walls of the tubes, the associated light source 34, 35 being triggered for each image or burst of images.

[0076] Figure 5 shows in more detail the cameras 30, 32 and light sources 34, 35 relative to the tubes 10. In this example, the cameras 30, 32 and light sources 34, 35 are shown moving in the x-axis relative to the tubes 10. As each camera 30, 32 passes over the centre of a tube 10, the corresponding light source 34, 35 is triggered, and an image or burst of images is taken.

[0077] Figure 6 shows schematically a top view of the tube inspection system. Movement of the cameras 30, 32 over the tubes 10 is shown diagrammatically with lines 44. The axes 42, 43 the cameras are aligned with the centres of the tubes as they pass overhead. The movement may be continuous, with the light sources triggered and images captured as the cameras move over the centres of the tubes.

[0078] Figure 7 shows parts of a tube inspection system in an embodiment of the invention. Referring to Figure 7, the system comprises first camera 30, second camera 32, first lighting source 34, second lighting source 35, infeed conveyer 36, datum stop 38, servomotor driven actuators 40, user interface 46, control unit 48, image processing unit 50 and alarm unit 52. The user interface 46 comprises input and output devices such as a display, keyboard, mouse etc. The control unit 48 and image processing unit 50 may be implemented as one or more software modules executing on a processor. As one example, the control unit 48 and imaging processing unit 50 may be implemented on a personal computer, although other suitable processing devices such as a microcontroller could be used as well or instead. The other parts of Figure 7 may be as described above with reference to Figures 4 to 6.

[0079] In operation, the user interface 46 is used to input commands to and receive results from the control unit 48 and the imaging processing unit 50. The control unit 48 controls the infeed conveyor 36 to feed a box of tubes into the gantry 26. When the box of tubes reaches the datum stop 38, the infeed conveyor 36 is stopped. The datum stop 38 is used as a reference to define the position of the box with respect to the inspection system. The control unit 48 then controls the actuators 40 to move the cameras 30, 32 and lights 34, 35 over the box of tubes. The movement is such that the cameras 30, 32 move over the centres of the tubes. This is possible because the packing configuration of the tubes is known as is the diameter of the tube (being entered into and / or stored by the control unit). Thus, with the box referenced at the datum stop 38 and the x-y movement of the cameras 30, 32 and lights 34, 35 under control either through positioning encoders or motor control, the control unit 48 can predicte when each camera 30, 32 is over the centre of each tube.

[0080] When the first camera 30 is over a tube, the first light source 34 is triggered and at the same time the camera 30 is used to take an image, or burst of images, of the inside of the tube. The camera 30 is focused on the bottom of the tube (the shoulder, nozzle and / or cap), although other parts of the inside of the tube may also be in its field of view. In this way, as the camera 30 moves over the tubes, a series of images is acquired of the insides of successive tubes, focused on the shoulder / bottom of the tubes.

[0081] Similarly, when the second camera 32 is over a tube, the second light source 35 is triggered and the camera 32 is used to take an image, or burst of images, of the inside of the tube. The second camera 32 has a larger depth of field than the first camera, so that most or a substantial part of the inside of the tube is in focus. The light source 35 emits a relatively high intensity light which can allow a high- quality image of the whole of the inside of the wall of the tube to be produced. As the camera 32 moves over the tubes, a series of images is acquired of the insides of successive tubes.

[0082] In practice, there may be some discrepancies between the predicted and actual positions of the cameras relative to the tubes. To compensate for this, in one embodiment, each camera 30, 32 is arranged to acquire a burst of 3 to 5 images of each tube as the camera passes over the tube (or some other number). The best image, or set of images, can then be selected for analysis. The images from the cameras 30, 32 are fed to the imaging processing unit 50 for analysis. The image processing unit 50 contains image processing algorithms which can process the image to identify edges and other features in the image. This typically involves comparing contrast levels in the image and using statistical considerations to determine when significant variations are present. This can allow defects such as faults in the manufacturing process or contamination to be identified. The defects are classified for accept, reject or further analysis. In the case of reject, a reject signal is output to the alarm unit 52. The alarm unit 52 outputs a visual or audible warning to the operator that a defect has been identified. If desired, this could be achieved via the user interface 46 as well or instead. Furthermore, the reject signal may be communicated to another apparatus, such as the tube manufacturing apparatus. This may allow production to be halted until the cause of the fault has been identified.

[0083] Figure 8 shows parts of the parts of the image processing unit 50 in more detail. Referring to Figure 8, the imaging processing unit comprises first image processing module 54, second image processing module 56 and classification module 58. The first image processing module 54 contains image processing algorithms which are used to process the images from the first camera 30 to identify potential defects. The second image processing module 56 contains image processing algorithms which are used to process images from the second camera 32 to identify potential defects. The categorisation module 58 is arranged to classify each tube as accept, reject or for further examination based on the potential faults identified by the first and second image processing modules 54, 56.

[0084] The images from the first camera 30 are passed to the first imaging processing module 54. When a burst of images from the first camera 30 is received, the image processing algorithms in the first imaging processing module 54 first identify the tube in each image, and then select the best image for further processing. This may be done, for example, by choosing an image in which the whole of the tube is in the image and / or based on other factors such as the quality of the image. Once an image has been selected, the image processing algorithms then process that image to identify the rim of the tube. The imaging processing algorithms then scan in decreasing circles concentric with the rim. Any differentials are recorded to build up a picture of significant grayscale changes. Features such as orifices and shoulder welds can be recognised because they are regular and known in advance, and can thus be eliminated from being reported as defects. This may be achieved by applying a mask to remove certain areas from the processing. For the bottom of the tube analysis, recognition of the shoulder enables the sidewall to be ignored. Classical software algorithms can determine the ovality of the tube, how far the centre of the tube is from its nominal position, if a cap is fitted, if a protective film is in place, if the shoulder of the tube is welded in place correctly, if there is foreign matter accumulated on the inner shoulder of the tube, etc. This provides mostly information about the process of manufacture and the packing of the tube. This information is passed to the classification module 58.

[0085] The images from the second camera 32 are passed to the second imaging processing module 56. When a burst of images from the second camera 32 is received, the image processing algorithms in the second imaging processing module 56 first identify the tube in each image, and then select the best image for further processing. The image processing algorithms then process the selected image to identify the rim of the tube in the image. The imaging processing algorithms then scan in decreasing circles concentric with the rim. Any differentials are recorded to build up a picture of significant grayscale changes. Features such as seams and shoulder welds are recognised because they are regular and can thus be eliminated from being reported as defects. Any unexpected particles which are identified in the image are likely to be caused by contamination. The contamination may be caused by the process of tube construction but is more likely to be random rather than systematic. The size of each particle can be estimated through spatial calibration of the pin hole camera image. The position of the particle is measured relative to another part of the tube such as the rim and / or the tube centre. The number of particles and their positions and sizes, together with any other defects, are passed to the categorisation module 58. The categorisation module 58 receives information concerning potential defects from the first image processing module 54 and the second image processing module 56, and uses this information to categorise the tube as accept, reject or for further examination. Certain faults, such as missing tube, missing cap, missing film, failed joint or significant debris may be automatically classified as rejection. Other defects may be classified as rejection only if multiple defects are identified and / or in dependence on their position and / or size. For example, where contamination is present, the tube may only be classified as rejection if the contamination is in a certain position on the tubular product and / or above a certain size. It is possible to combine a number of contamination defects within the classification such that multiple small defects which would normally be acceptable individually result in a tube reject. These thresholds can be set via the user interface 46 to suit the manufacturing operation being undertaken.

[0086] A typical image from the first camera 30 is shown in Figure 9. The area of interest of the first camera is the bottom of the tube. This area can be identified automatically by the image processing algorithms, by identifying the boundary between the sidewall 22 and the shoulder 14, and setting the area of interest to be within the boundary. Once the image has been processed, it can be annotated to highlight potential defects, for example, using a colour such as red.

[0087] A typical image from the second camera 32 is shown in Figure 10. In this example, the image of the tube has been taken with the axis of the camera at an angle to the longitudinal axis of the tube. The area of interest of the second camera is the sidewall 22. This area can be identified automatically by identifying the area between the rim 21 and the shoulder 14, and setting the area of interest to be between the two. The image has been annotated to highlight potential defects. The annotated images can be passed to the user interface 46 for display and / or to maintain a record of the inspection process.

[0088] In the arrangement described above, the first camera 30 has a relatively small depth of field, and is focused on the closed end of the tube (the shoulder end). This can allow the closed end of the tube to be imaged with a high level of detail. In practice, many of the faults which may occur in the manufacture of a tube, such as missing cap, missing film, incorrect connection of the shoulder to the sleeve, incorrect formation of the orifice and debris on the shoulder, are likely to be found at the end of the tube. Thus, focusing the first camera on the end of the tube can facilitate the identification of these faults (which are likely to be production faults) in the image. However, since the sleeve of the tube will for the most part be out of focus, the imaging software will be less able to identify any faults (such as contamination) in the sleeve.

[0089] On the other hand, the second camera 32 is of a pin-hole type with a large depth of field. This allows substantially the entire inner wall of the tube to be substantially in focus. Furthermore, a high intensity light source is used to allow the second camera to capture an image without the need for a long exposure time. Although the level of detail in the images produced by the second camera may be less than that in the images produced by the first camera (due to the larger depth of field), it has been found that use of a second camera with a large depth of field can allow contamination and other faults to be identified along the length of the tube with a high degree of reliability.

[0090] The use of a high intensity light source 35 with the second camera 32 allows the sidewall 22 to be imaged with sufficiently high quality, using a large depth of field and without extending exposure times. However, the high intensity lighting may produce reflections from the bottom of the tube, which may reduce the ability of the image processing algorithms to detect features in the bottom of the tube. In this case, limiting the area of interest in the second image to the sidewall can help to reduce or avoid any reflections from the bottom of the tube from interfering with the image analysis. On the other hand, the intensity of the first light source 34 is chosen so as not to produce significant reflections from the bottom of the tube. This can allow the first camera 30 to image the bottom of the tube without reflections which would interfere with the image analysis.

[0091] Furthermore, by classifying the tube for accept, reject or further examination based on images from both the first and second camera, it may be possible to perform the classification with a greater degree of reliability than would otherwise be the case. For example, a potential fault may not be sufficient to reject the tube if it can only be identified from one image, but may be sufficient to reject the tube if it can be identified from both images. Furthermore, an accumulation of potential faults (such as contamination) in one or both images may lead to a rejection even if the individual faults on their own would not have done.

[0092] If desired, polarising filters could be added to the first camera and / or the second camera to minimise the impact of internal reflection within the tubes. This may be of relevance where there are known to be metallic internal surfaces, such as with aluminium tubes, or where there is a metallised tamper evident seal at the end of the tube.

[0093] The embodiments of the tube inspection system described above are given by way of example only, and various modifications could be made. For example, the first camera need not have a fixed focal length lens and may have any other suitable type of lens (for example, telecentric). The second camera need not be of the pin-hole type, and may be of any type with a depth of field greater than the first camera. The first and second camera need not be adjacent to each other, and need not face in the same direction. If desired, the system could be adapted such that the cameras and light sources are stationary and the tubes move in front of the imaging system. This could be achieved, for example, by passing the tubes in front of the cameras on a conveyor. Alternatively, it would be possible for both the cameras / light sources and the tubes to be moved.

[0094] Figure 11 shows parts of a tube inspection system in another embodiment of the invention. Referring to Figure 11 , the system comprises first camera 30, first light source 34, second camera 32a, second light source 35a, third camera 32b and third light source 35b. Other parts of the tube inspection system may be substantially the same as in the embodiments described above.

[0095] In this embodiment, the first camera 30 is focused on the bottom of the tubes 10. Thus, the first camera and first light source may be substantially the same as the first camera and first light source described above. The second and third cameras 32a, 32b are focused on the sidewalls. Thus, the second and third cameras 32a, 32b and the second and third light sources 35a, 35b may be substantially the same as the second camera 32 and the second light source 35 described above. However, in this embodiment, the second and third cameras 32a, 32b are tilted so that their lines of sight are at an angle to the longitudinal axes of the tubes. The second camera 32a is angled in one direction and the third camera 32b is angled in the opposite direction. The amount of angle is chosen so that the camera is able to capture an image of at least a majority of the sidewall between the rim and the shoulder, in at least one part of the tube circumferentially. By using two cameras at different angles to capture images of the sidewalls, it may be possible to obtain better quality images of the sidewalls.

[0096] In another arrangement, four cameras could be used to capture images of the sidewalls, with each camera angled towards one of the four quadrants of the inside of the tube. In general, any suitable number of cameras may be provided, each of which may be parallel to or at any suitable angle to the to the longitudinal axes of the tubes.

[0097] In another embodiment, rather than tilting the camera, the Scheimpflug principle could be used to adjust the tilt of the focal plane with respect to the camera sensor plane. This can be achieved by tilting the camera lens with respect to the sensor plane or by using a Scheimpflug adaptor.

[0098] In a further embodiment of the invention, an auxiliary camera (or one of the first and second cameras) is used to take an image of a box of tubes, or at least part of the box of tubes, before scanning begins. The image from the auxiliary camera is then used to identify the type of tube and / or the packing configuration. Furthermore, the image from the auxiliary camera can be used by the control unit 48 to ensure that the cameras pass over the centres of the tubes, and to ensure that the light sources and cameras are triggered at the correct moments. This may help to improve accuracy and may avoid the need to take multiple images of each tube.

[0099] The embodiments described above may provide the following technical aspects:

[0100] • The speed at which inspection takes place may allow tube manufacturing speeds of up to / in excess of, for example, 240 tubes per minute to be 100% inspected. • The full inspection of the inside of a tube allowing defects to be identified and contamination of components below, for example, 1 mm2to be flagged.

[0101] • The use of a classification system based on particle count, position and size to determine if a tube is to be accepted or rejected.

[0102] • The use of two camera and a pulsed lighting system. This allows simpler less expensive cameras to be used to capture high quality, high definition images of the inside of a tube that facilitate discrimination when using computer algorithms.

[0103] • The continuous movement of the cameras to generate a series of images based on the packing pattern of the tubes, the known size of the box, registration of the box at a datum and the diameter of the tubes packed into the box.

[0104] • The intense lighting for the pin hole camera that can fully utilise the large depth of field of the camera without extending exposure times.

[0105] • The optional addition of optical filters that allow for the suppression of unwanted reflections or enhance features of interest.

[0106] The system, although designed for tubes used in the cosmetic, consumer products and pharmaceutical industries (laminate, paper, plastic and metal), would be suitable for any container with a large aspect ratio of depth to diameter that is required to be 100% inspected for defects and / or contamination.

[0107] Figure 12 shows steps taken by the control unit to scan the cameras over the tubes in one embodiment. Referring to Figure 12, processing starts in step 100. In step 102, the infeed conveyor is activated, and a box of tubes is moved into the gantry. In step 104 it is determined whether the box has reached the datum stop. If the box has not reached the datum stop, then processing returns to step 102 and the box continues to be moved into the gantry. If the box has reached the datum stop, then the infeed conveyor is stopped, and processing proceeds to step 106. In step 106 the tube type and packing configuration is acquired. This may be done, for example, by entering the appropriate tube type and packing configuration via the user interface. Alternatively, it may be possible to determine the tube type and / or packing configuration automatically, for example by scanning a barcode or other feature of the box, or for the tube type and / or packing configuration to be communicated from another piece of apparatus such as a tube making machine, or by analysing an image of the box or part of the box. Where a plurality of boxes of the same type are to be inspected, it may be assumed that the tube type and packing configuration is the same as for the previous box. Optionally, in step 108 settings of the first and / or second cameras are adjusted. For example, the focus of the first camera and / or the second camera, and potentially the intensity of the first and / or second light source, and / or the depths of field and / or the exposure times, may be adjusted in dependence on the type of tube.

[0108] In step 110 the control system uses the motion track to move the first camera and first light source to the predicted centre of the next tube (or the first tube in the case of the first iteration). This is possible because the box is referenced at the datum, and the packing configuration and diameter of the tubes is known. In step 112 the first light source is activated. This causes the inside of the tube to be lit. At the same time, in step 114, the first camera is used to capture an image of the inside of the tube. The first camera is focused so that the image is focused on the tube shoulder and cap. If desired, a burst of two or more images may be captured when the first camera is predicted to be over the tube. This can allow the best image to be selected at a later stage.

[0109] In step 116, the control unit uses the motion track to move the second camera and second light source to the predicted centre of the tube (which may be the same as or different from that which was imaged by the first camera). In step 118 the second light source is activated. This causes the inside of the tube to be lit. At the same time, in step 120, the second camera is used to capture an image of the inside of the tube. The second camera has a relatively large depth of field, so that the whole of the inside wall of the tube is substantially in focus. The second light source has a relatively high intensity, to ensure that images of sufficiently high quality are obtained without the need for extended exposure times. If desired, a burst of two or more images may be captured when the second camera is predicted to be over the centre of the tube.

[0110] In step 122 it is determined whether images have been taken of all of the tubes in the box. If not all of the tubes have been imaged, then processing returns to step 110 and the first camera is moved to the next tube. If all of the tubes have been imaged, then processing ends in step 124.

[0111] Figure 13 shows steps taken by the first image processing module in one embodiment. Referring to Figure 13, in step 130 a burst of images is received from the first camera. In step 132, the images are analysed to identify the location of the tube (if present) in the images. In step 134 it is determined whether the tube is missing or misplaced. If the tube is missing or misplaced, then in step 136 a missing / misplace tube error is recorded, and processing then proceeds to step 162. On the other hand, if it is determined in step 134 that the tube is not missing or misplaced, then processing proceeds to step 138.

[0112] In step 138, the best image from the burst of images is selected. For example, the image in which the tube is most central may be selected, although the selection may be based on other factors, such as image quality, as well or instead. Furthermore, if desired, more than one image could be selected for analysis. In step 140, the selected image is scanned to locate the rim of the tube. In step 142, the ovality of the tube is determined. This may be achieved using classical image processing algorithms, which can measure the distances between edges in the image. Optionally, the distance of the centre of the tube from its nominal position is determined by measuring the appropriate distance in the image.

[0113] In step 144, the boundary between the shoulder and the sidewall is identified. In step 146, the area of interest in the first image is set. This may be the area inside the boundary between the shoulder and the sidewall, which will generally correspond to the bottom of the tube, although some or all of the sidewall could be included if desired. In step 148 a mask is optionally applied to exclude any features which are not to be taken into account in the analysis.

[0114] In step 150 the area of interest is inspected. In one embodiment, this is achieved by applying a variance analysis algorithm to the relevant part of the image. This may involve, for example, finding the tube centre and then detecting anomalies in the grey levels on concentric profiles. Other algorithms such as local thresholding (searching for objects darker than the local background) and / or difference to a median model (comparison with a model generated from images of a number of tubes which are known to be good) may be used as well or instead. All of these techniques may use parameters and thresholds which may be variable to give the user the option of tuning the inspection.

[0115] In step 152, it is determined whether a cap is fitted. This can be done by comparing the intensity of the image at the position where the cap is expected to that which would be expected if the cap was present (or was not present). In step 154, it is determined whether a protective film is in place. This can be done by comparing the intensity of the image at the position where the film is expected to that which would be expected if the film was present (or was not present). In step 156, it is determined whether the shoulder of the tube is correctly attached to the tube sleeve. This can be done by comparing features of the shoulder to those which would be expected if the shoulder was corrected attached (or to a known fault). In step 158, it is determined whether any foreign matter is present on the inner shoulder of the tube. This can be done by identifying any unexpected intensity differentials in the image, and comparing their size to a threshold. As part of this step, the size and position of the particles may be recorded, together with the number of particles. If desired, any other parameters of the tube and / or potential faults could be measured or identified as well or instead.

[0116] In step 160, the image of the tube is annotated to highlight any potential foreign matter or other faults identified in steps 134 to 158. This may be done by marking the potential faults on the image in a colour such as red, or in any other way. The annotated image may then be provided to the user interface for display. In step 162, the characteristics of the end of the tube determined in steps 134 to 158 are reported to the classification unit. Processing can then proceed to the image or burst of images from the first camera (step 130).

[0117] Figure 14 shows steps taken by the second image processing module in one embodiment. Referring to Figure 14, in step 170 a burst of images is received from the second camera. In step 172, the images are analysed to identify the location of the tube (if present) in the images. If desired, it may be determined whether the tube is missing or misplaced, in a similar way to that shown in Figure 13. In step 174, the best image from the burst of images is selected. For example, the image in which the tube is most central may be selected, although the selection may be based on other factors, such as image quality, as well or instead. Furthermore, if desired, more than one image could be selected for analysis. In step 176, the selected image is scanned to locate the rim of the tube. In step 178, the boundary between the shoulder and the sidewall is identified. In step 180, the area of interest in the first image is set. This may be the area between the rim, and the boundary between the shoulder and the sidewall, which will generally correspond to the sidewall of the tube. However, some or all of the end of the tube could be included if desired. Optionally, a mask is applied to exclude any features which are not to be taken into account in the analysis.

[0118] In step 182 the area of interest is inspected. In one embodiment, this is achieved by applying a variance analysis algorithm to the relevant part of the image. This may involve, for example, unrolling the image between the rim and the shoulder, and then detecting anomalies in the grey levels in the unrolled image. Other algorithms such as local thresholding and / or difference to a median model may be used as well or instead. All of these techniques may use parameters and thresholds which may be variable to give the user the option of tuning the inspection.

[0119] In step 184, it is determined whether any faults are present in the sidewall. This may be done by comparing features in the image to those which would be expected if the tube was correctly formed. In step 186, it is determined whether any particles are present on the inner wall of the tube. This can be done by identifying any unexpected intensity differentials in the image. If no particles are present, then processing proceeds to step 192. If particles are present, then in step 188 the sizes and positions of the particles are measured. In step 190 the number of particles together with their sizes and positions are recorded. If desired, other characteristics of the tube could be identified as well or instead. In step 192, the image of the tube is annotated to highlight any potential foreign matter or other faults. This may be done by marking the potential faults on the image in a colour such as red, or in any other way. The annotated image may then be provided to the user interface for display. In step 194, the characteristics of the inner wall of the tube determined in steps 184 to 190 are reported to the classification unit. Processing can then proceed to the next image from the second camera (step 170).

[0120] It will be appreciated that the steps shown in Figures 13 and 14 are given by way of example only. Some of the steps may be omitted, and other steps may be carried out as well or instead. The steps may be carried out in a different order from that shown. Steps carried out by the first image processing module may also be carried out by the second image processing module, and vice versa. Furthermore, the characteristics of the tube are given by way of example only, and other features of the tube could be determined instead or as well. For example, the characteristics to be determined could be input via the user interface, and / or set depending for example on the type of tube or manufacturing process.

[0121] Figure 15 shows steps carried out by the classification unit in one embodiment. Referring to Figure 15, in step 200 the characteristics of the end of the tube are received from the first image processing unit. In step 202, the characteristics of the inner wall of the tube are received from the second image processing unit. In step 204, the various characteristics are compared to thresholds. In step 206, the tube is classified as accept, reject or for further examination based on the results of the comparisons. For example, certain defects (such as missing tube or missing parts) may cause classification as reject. Other defects (such as spots in the image) may cause classification as accept or for further examination if they are few in number, or less than a certain size, or in a less critical area. On the other hand, the defects may cause classification as reject if their number and / or size exceeds a certain threshold and / or if they are in a more critical position. In step 208 the classification of the tube is output to the user interface. In step 210 it is determined whether a defect has been classified as reject. If no defect has been classified as a fail, then processing of that tube stops in step 214. If a defect, or combination of defects, has been classified as reject, then in step 212 a reject signal is produced. The reject signal may be output to the alarm unit, to the user interface and / or to another piece of apparatus such as a tube making machine. Processing of that tube then stops in step 214. It will be appreciated that the steps shown in Figure 15 are given by way of example, and the steps could be carried out in a different order and / or other steps carried out instead or as well.

[0122] In alternative embodiments, other techniques for analysing the images and classifying the tubes may be used instead of or as well as those described above. For example, in one alternative embodiment, artificial intelligence (Al) tools could be used to analyse the images and classify the tubes. This would typically involve using machine learning to train a model using training data from tubes with known faults. The trained model can then be used to classify tubes based on actual image data.

[0123] Embodiments of the invention may provide some or all of the following:

[0124] • A system for the inspection of internal surfaces of tubes or other similar product containers where the length to diameter ratio is typically between 5 and 8 (e.g. within the size ranges of 28mm to 40mm diameter and maximum length of 220mm).

[0125] • An inspection system based upon vision that requires a high definition image that enables the deployment of mathematical models to distinguish contamination from background.

[0126] • A system that can estimate the size of contamination and so classify the contamination for use in rejection or acceptance criteria that may be user configurable.

[0127] • The use of two collocated cameras that have different focal lengths. The first being of a general type that can image the furthermost internal surface of the tube and is used for contamination on the shoulder, shoulder defects, cap defects and film defects. The second camera having a large depth of field forming an image when an intense light is provided, of the whole inner surface of the tube, the image being used for examining defects in the seam of laminate tubes and for contamination on the inner surface of all tube types.

[0128] • The synchronisation of a high intensity burst of light with the triggering of the pin hole camera to collect a set of images as it travels along the calculated centre line of each row of tubes in a box to be inspected. • The movement of the cameras and lighting over the tubes packed in boxes, the image capture being based on the expected tube rows position.

[0129] • Alternatively the movement of the tubes in a controlled way in front of the camera system.

[0130] • A tube inspection system that derives information based upon the expected position of a tube with respect to a camera position.

[0131] • A tube inspection system that derives information based upon the expected tube position such that images sufficiently concentric to the tube centre to be useful for inspection are identified.

[0132] • A tube inspection system where the optical image of a tube is enhanced or improved through the deployment of optical filters on the camera systems used.

[0133] Preferred features of the invention have been described above with reference to various embodiments. However, the invention is not limited to these embodiments, and variations in detail may be made. For example, the system could be adapted for use with any packaging or tubular product where there is the need to look at the inside of the package walls and at the inside end of the package e.g. ampules or vials. Various other modifications will be apparent to the skilled person within the scope of the appended claims.

Claims

CLAIMS1 . Apparatus for inspecting a plurality of tubular products in a container, the apparatus comprising: a first camera arranged to capture a first image of an inside of a tubular product with a first part of the tubular product substantially in focus; a second camera arranged to capture a second image of the inside of the tubular product with a second part of the tubular product, different from the first part, substantially in focus; and means for analysing the first and second images to determine whether there is a defect in the tubular product.

2. Apparatus according to claim 1 , wherein the tubular product comprises one end which is at least partially open and another end which is at least partially closed.

3. Apparatus according to claim 1 or 2, wherein the first camera is focused on an at least partially closed end of the tubular product.

4. Apparatus according to any of the preceding claims, wherein the second camera is focused on an inner wall of the tubular product.

5. Apparatus according to any of the preceding claims, wherein the second camera has a different depth of field from the first camera.

6. Apparatus according to any of the preceding claims, wherein the second camera has a depth of field such that at least a majority of an inner wall of the tubular product is substantially in focus.

7. Apparatus according to any of the preceding claims, the apparatus further comprising: means for providing relative movement between the cameras and the tubular products; and means for controlling the relative movement to bring the tubular products into a field of view of the cameras.

8. Apparatus according to claim 7, wherein the means for providing relative movement is arranged to provide continuous movement between the cameras and the plurality of tubular products, and the first and second cameras are arranged to capture the first and second images during the continuous movement.

9. Apparatus according to claim 7 or 8, wherein the plurality of tubular products are provided in a container with a predetermined packing configuration, and the control means is arranged to control relative movement between the cameras and the tubular products in accordance with the predetermined packing configuration.

10. Apparatus according to any of claims 7 to 9, further comprising means for conveying the container into the inspection apparatus.11 . Apparatus according to any of claims 7 to 10, further comprising a datum stop arranged to locate the container relative to the apparatus, wherein the control means is arranged to move the cameras relative to the tubular products using the datum stop as a reference.

12. Apparatus according to any of claims 7 to 11 , wherein the tubular products have a predetermined diameter, and the control means is arranged to move the cameras relative to the tubular products in accordance with the predetermined diameter.

13. Apparatus according to any of claims 7 to 12, wherein the control means is arranged to trigger the first camera to take the first image when it is predicted that the tubular product is in a field of view of the first camera, and to trigger the second camera to take the second image when it is predicted that the tubular product is in a field of view of the second camera.

14. Apparatus according to claim 13, wherein at least one of the first and second cameras is arranged to capture a plurality of images of the tubularproduct when it is triggered, and the analysing means is arranged to select at least one of the plurality of images for analysis.

15. Apparatus according to any of the preceding claims, further comprising a first light source arranged to at least partially illuminate a field of view of the first camera and / or a second light source arranged to at least partially illuminate a field of view of the second camera.

16. Apparatus according to claim 15, wherein the first light source is triggered when an image is taken by the first camera and / or the second light source is triggered when an image is taken by the second camera.

17. Apparatus according to claim 15 or 16, wherein the first light source has a first intensity and the second light source has a second intensity different from the first intensity.

18. Apparatus according to any of the preceding claims, wherein the defect comprises one or more of: a manufacturing fault; a packing fault; and contamination in the tubular product.

19. Apparatus according to any of the preceding claims, wherein the analysing means is arranged to analyse the first image to determine whether there is a defect in an at least partially closed end of the tubular product.

20. Apparatus according to any of the preceding claims, wherein the analysing means is arranged to analyse the second image to determine whether there is a defect in an inner wall of the tubular product.21 . Apparatus according to any of the preceding claims, wherein the analysing means is arranged to analyse the first image and / or the second image to determine a physical property of the tubular product.

22. Apparatus according to any of the preceding claims, further comprising means for classifying the tubular product as accept or reject in dependence on anoutput of the analysing means, and optionally means for producing a reject signal if the tubular product is classified as reject.

23. Apparatus according to any of the preceding claims, wherein the tubular product is a collapsible tube with one end at least partially open prior to filling and / or of a type which is arranged to be filled with a dispensable product.

24. A method of inspecting a plurality of tubular products in a container, the method comprising: capturing a first image of an inside of the tubular product with a first part of the tubular product substantially in focus; capturing a second image of the inside of the tubular product with a second part of the tubular product, different from the first part, substantially in focus; and analysing the first and second images to determine whether there is a defect in the tubular product.

25. Apparatus for inspecting a plurality of tubular products, the apparatus comprising: at least one camera arranged to capture images of the tubular products; means for providing relative movement between the camera and the tubular products; means for controlling the relative movement to bring the tubular products into a field of view of the camera; and means for analysing the captured images to determine whether there is a defect in the tubular product, wherein the plurality of tubular products are provided in a container with a predetermined packing configuration, and the control means is arranged to control the relative movement in accordance with the predetermined packing configuration.

Images