Apparatus and method for inspecting a container that is at least partially transparent to a given electromagnetic radiation.

Abstract

The invention relates to a device 4 for inspecting a container 8 containing a liquid at least partially transparent to a given electromagnetic radiation, the container 8 comprising at least one portion with radial symmetry about an axis of symmetry X, the device 4 comprising: - a rotation device 5 adapted to rotate the container 8 about an axis of symmetry X, a camera 12 arranged so as to be located within the field of view of the container 8 when it is placed therein, the camera 12 being sensitive to predetermined electromagnetic radiation; a processing unit 32 adapted to control the rotation device 5 and the camera 12, controlling the rotation device 5 to move the container 8 at a first angular velocity vmax1 and to keep the first angular velocity constant over a first time period t1; controlling the camera 12 to acquire at least a first and a second series of images while rotating at a first constant angular velocity vmax1, each image 40 of the first or second series being an image of a portion 16 of the container 8, each series of images representing a 360° rotation of the portion of the container; identifying defect areas 60 in a first and second series S1, S2 of images 40, each defect area having at least one characteristic different from the characteristics of adjacent areas, and generating first and second maps S1, S2 of the defect areas, each map S1, S2 comprising the locations and characteristics of the defect areas, wherein identical locations in the first map and the second map identify the same locations in the container 8; Comparing the locations of the defect areas in the first and second maps S1, S2, If a defect area 60 is present at a location in the first map S1 and a defect area is present in an area surrounding the same location in the second map S2, it establishes that a first impurity 70 is present in the container 8 or in the liquid contained in the container. A processing unit 32 programmed to Equipped with.

Description

[Technical Field]

[0001] The present invention relates to an apparatus and a method for inspecting an at least partially transparent container using a camera. [Background Art]

[0002] Analysis of medical substance containers to detect impurities inside the container itself is essential, not only in the medical field. In fact, since it is unacceptable that a medical substance contains any type of contaminant, if any impurities are present, the container must be discarded.

[0003] Systems adopted in the art are optical, using a camera that scans each container to detect contaminants.

[0004] Known optical systems generally detect the presence of contaminants / impurities by inspecting images generated by a camera. The approach used by the applicant himself is for example of the "spin-and-stop" type, i.e., before the container reaches the front of a particle analysis camera, the container is rotated about its own axis. The container is then stopped, the liquid inside the container continues to move due to inertia, while the remaining part of the container is stationary, and the camera detects contaminants in the liquid while the liquid is still moving.

[0005] However, this system is not effective when there are air bubbles in the liquid. Even when the container is stopped, air bubbles move similarly to contaminants, therefore, although the above-mentioned "spin-and-stop" approach can identify whether "something is present in the liquid" inside the container, it cannot identify whether the something is an internal contaminant or an air bubble.

[0006] Therefore, when a liquid that can form bubbles is present, analyzing a container containing such a liquid to look for contaminants will result in numerous false positives, and in some cases, the contaminants are not actually present, leading to either excessive and improper disposal of containers or double-testing of discarded containers. [Overview of the project] [Problems that the invention aims to solve]

[0007] It is desirable to make available methods and apparatus for inspecting at least partially permeable cylindrical containers that are adapted to at least partially identify whether contaminants are present in the liquid inside the container and to distinguish them from any bubbles that may form in the liquid. [Means for solving the problem]

[0008] According to one aspect, the present invention relates to an apparatus for inspecting a container containing a liquid that is at least partially transparent to a predetermined electromagnetic radiation, wherein the container comprises at least one part having radial symmetry about an axis of symmetry, and the apparatus is - A rotating device adapted to rotate the container around an axis of symmetry, - A camera positioned such that the container is within its field of view, the camera being sensitive to the predetermined electromagnetic radiation, - A processing unit adapted to control a rotating device and a camera, The container is moved at a first angular velocity, and the rotating device is controlled to maintain the first angular velocity constant over a first period of time. The camera is controlled to acquire at least a first and second series of images while rotating at a first constant angular velocity, where each of the first or second series of images is an image of a portion of the container, and each series of images represents a portion of the container rotated 360°. • In a series of first and second images, defect regions are identified, each defect region having at least one characteristic different from the characteristics of adjacent regions, first and second maps of the defect regions are generated, each map comprising the location and characteristics of the defect region, and identical locations in the first and second maps identify the same location within the container. Compare the locations of the defect areas in the first and second maps. If a defect region exists at a location in the first map, and another defect region exists in the area surrounding the same location in the second map, then it is proven that the first impurity is present in the container or in the liquid contained in the container. A processing unit programmed in such a way It is equipped with.

[0009] In a further embodiment, the present invention is The step of providing a container that is at least partially transparent to a predetermined electromagnetic radiation and contains a liquid that is at least partially transparent to a predetermined electromagnetic radiation, wherein the container comprises at least one part having radial symmetry about an axis of symmetry, The steps include rotating the container at a first angular velocity and maintaining the first angular velocity constant over a first period of time, The steps include: acquiring at least a first and a second series of images during rotation at a first constant angular velocity, wherein each of the first and second series of images is an image of a portion of the container, and each series of images represents the portion of the container rotated 360°; The steps include: identifying defective regions in a series of first and second images, each defective region having at least one characteristic different from the characteristics of adjacent regions, and generating first and second maps of the defective regions, each map comprising the location and characteristics of the defective region, wherein identical locations in the first and second maps identify the same location within the container; - A step of comparing the locations of defect regions in the first and second maps, - If a defect region exists at a location in the first map, and a defect region exists in the area surrounding the same location in the second map, the step is to prove that the first impurity is present in the container or in the liquid contained in the container. This relates to methods for inspecting containers, including [specific details omitted].

[0010] In the present invention, the container is preferably inspected for medical use, but is not limited thereto. The container may be, for example, a vial, a bottle, or an ampoule. The container comprises a hollow body defining the side wall. These may also have a cap for closing the hollow body, a cap that can be removed when access to the internal liquid is needed. Alternatively, the cap may be punctured with a syringe. The side wall is partially transparent to a given electromagnetic radiation, at least in part thereof. Preferably, the side wall is transparent to a given electromagnetic radiation. Thus, “a partially transparent or transparent container” means a container having at least a portion of its side wall that is at least partially transparent or transparent.

[0011] The liquid is contained within the hollow body. The liquid is also at least partially permeable to a given electromagnetic radiation. Preferably, the liquid is permeable to a given electromagnetic radiation.

[0012] The liquid may be, for example, a drug, or distilled water combined with a drug. The liquid may have varying densities and some viscosity. The critical density is the density at which the liquid does not move relative to the container when the container is rotated. A characteristic of the liquid is its tendency to form bubbles. Bubbles are small amounts of air trapped within and surrounded by the liquid.

[0013] It should be noted that the container is permeable, and as a result, the permeable or semi-permeable liquid placed inside the container can be seen from the outside. Therefore, the container is preferably made of glass or plastic, such as plexiglass or a cyclic olefin copolymer (COC) type copolymer.

[0014] Furthermore, the container has a portion that exhibits radial symmetry. Preferably, with respect to an axis of symmetry hereafter referred to as axis X, the hollow body comprises a rotating solid around this axis. For example, the rotating solid may be a cylinder. However, other solids exhibiting radial symmetry may be used. The entire container does not need to be radially symmetry; it is sufficient if a portion large enough to accommodate all the liquid placed in the container has such radial symmetry.

[0015] The inspection is performed by the apparatus and / or method according to the present invention. An object of the present invention is, in at least some cases, to identify defects in a container and to distinguish bubbles from other existing defects.

[0016] To inspect a container, it is rotated around its (radial) axis of symmetry. The container can be rotated in various ways, for example, depending on its size and / or weight. For example, the container may be rotated by a rotating device, which is part of the apparatus of the present invention, and the rotating device may include a support on which the container rests. The support is then rotated. For example, the rotation may be controlled and actuated by a motor. The rotating device may include a grip to which a motor means for rotating around the axis of symmetry is attached. Preferably, the rotating device is positioned so as not to obstruct camera vision (as detailed below), and thus the checking of any defects that may occur inside the container. For this purpose, a rotating device that grips or otherwise pulls the rotating container from its support is preferred. There are many gripping devices / methods suitable for this purpose. For example, a gripper on the bottom head of the container or on the neck of the container itself can be gripped. Importantly, these devices rotate the container around the axis of symmetry.

[0017] The inspection of the container for defects is performed using a camera. The camera is preferably a linear camera, but may also be a 2D camera (in other words, capable of generating a two-dimensional image). The term "camera" also includes a photo camera. The camera is positioned so that the container being inspected is within its field of view. The entire container does not need to be within the camera's field of view; it is sufficient if only a "part" of the radially symmetrical components of the container is within the camera's field of view. The camera may, for example, face the container, particularly the side wall of the container, or the camera may be oriented differently, and one or more mirrors can be used to ensure that the container remains within the camera's field of view.

[0018] The camera is sensitive to electromagnetic radiation, and the side walls of the container and the liquid are transparent to the electromagnetic radiation. Preferably, the electromagnetic radiation is electromagnetic radiation in the visible range, that is, radiation with a wavelength of 390 nm to 700 nm. The electromagnetic radiation may also be included in the near-infrared (IR) spectrum, that is, radiation with a wavelength of 700 nm to 1 mm. That the camera is sensitive to electromagnetic radiation means that the camera comprises at least one sensor sensitive to such radiation. The sensor may be, for example, a CCD or a CMOS sensor.

[0019] When the container is placed on a rotating device, the container rotates about an axis of symmetry. Control for rotating the container can be provided, for example, by a processing unit that controls both the rotating device and the camera.

[0020] Next, the container is subjected to a first acceleration from a stopped state to a first predetermined angular velocity. The first predetermined angular velocity can be set, for example, via the processing unit. The predetermined angular velocity is variable and depends on the type of container, the type of liquid contained therein, and the type of impurities expected in the container. Preferably, the predetermined angular velocity for impurities such as glass, metal, rubber, plastic, fibers, for example plastic fibers (especially polyester or fibers used in making clothes), hair, etc., is 200 rpm to 10000 rpm, more preferably 500 rpm to 5000 rpm. For example, the predetermined angular velocity may be 2500 rpm.

[0021] Accordingly, the container goes through a first acceleration step from the stopped state to a rotating state at the first predetermined angular velocity. The acceleration step may last, for example, 50 milliseconds to 2 seconds, more preferably 200 to 700 milliseconds. The duration of acceleration depends on the size of the container (especially, its radius if the container has cylindrical symmetry), the weight of the liquid contained in the container, the volume of the liquid contained in the container, and the properties of the liquid contained in the container.

[0022] A predetermined angular velocity is kept substantially constant over a first period. During this first period, the camera is activated, for example via a processing unit, and as a result acquires images of portions of the container, wherever they are located within its field of view or wherever selected portions of the container are located, at angular intervals, preferably at constant angular intervals. Furthermore, the camera remains active, i.e., continues to acquire images for at least first and second full rotations of the container about the container's axis of symmetry. The first period must be long enough to allow the camera to acquire images of portions of the container that have radial symmetry with respect to a first and second full rotation of the container over 360°.

[0023] The first period is continuous (a single period). The first angular velocity is kept constant throughout the first period without any interruptions such as acceleration, deceleration, or stopping.

[0024] The camera can capture images of the container even when the container is not rotating at a first constant angular velocity, but these images are not part of the subsequent processing.

[0025] In each of the first and second rotations, the camera acquires M images. Preferably, images are acquired at angular intervals Δθ. Preferably, the angular resolution is at least 0.02 radians, more preferably at least 0.01 radians. The resolution may be in the range of 100 μm / pixel to 1 μm / pixel, and preferably 10 μm / pixel. In the image, the development of the side of the container is realized by a 360° rotation, so if the radius of the container is r, it is 2*π*r. To have a resolution of 10 μm, each angular interval Δθ (see the definition of Δθ below) is preferably equal to the following: 2π / (2*π*r) / 10μm radians

[0026] In other words, the expanded 2πr is divided into several N segments at 10 μm intervals. This number N is also preferably the number of intervals into which the inscribed angle is 360° (or 2π).

[0027] Each of the M images is, for example, a row of pixels (single-base pixels) in the case of a linear camera, or a frame of pixels in the case of a two-dimensional camera. Therefore, each image can be viewed as a grid with a base of only one pixel and a height of P pixels, or as an array of dimensions P × Q. The images are essentially raster images, and each element of the raster, called a pixel, is associated with a specific color or grayscale.

[0028] A grayscale image is an image in which each pixel takes a value within a range. Typical values ​​are [0,63], [0,255], and [0,1023] per pixel, with 6, 8, and 10 bits respectively.

[0029] Color can be defined using two techniques. If an image contains only a few colors (up to 256), a list of colors to be used is generated, and indices pointing to the color of a specific pixel are inserted into the raster. If the image contains many colors, a single pixel directly defines its color, rather than defining an index pointing to a color bullet. Color can be defined, for example, as a combination of three components: blue, red, and green (RGB system).

[0030] M images acquired in the first rotation and M images acquired in the second rotation form the first and second sets of images. These images are preferably associated with each other. This association is preferably performed by a processing unit. In the association, an aggregated image is generated for each set of M images. Thus, the association of the first set of M images generates the first aggregated image, and the association of the second set of images generates the second aggregated image. This association is performed seamlessly to obtain an aggregated image of the container unfolding, in other words, the aggregated image represents the outer sidewalls of the container over the entire 360° unfolding. The camera, together with the processing unit, generates a continuous unfolding of the container's sidewalls, formed by assembling M images of "slices" of the same container acquired at angular distances equal to Δθ from each other.

[0031] In the case of a linear camera, the images are easily combined sequentially without any other manipulation. In the case of a 2D camera, it is necessary to create an image overlay using appropriate software known in the field so that overlapping portions of the outer side walls of the container do not appear in the aggregated image.

[0032] For example, for a linear camera and a cylindrical container with a diameter of 24 mm, the number of images (M) is equal to 5000.

[0033] A linear camera acquires lines at constant angular intervals dθ while the container being inspected rotates around its own axis. For example, by using a rotary encoder connected to a motor responsible for the container's rotation, the camera generates a continuous unfolding of the container's sidewalls. The encoder can be controlled by a processing unit.

[0034] Each pixel in the aggregated image uniquely corresponds to a single point on the side wall of the container (excluding the line at the end of the image, as is known). Therefore, the coordinates of each pixel in the aggregated image correspond to an exact point within the side wall of the container. As a result, the position in the aggregated image corresponds to a position on the side wall of the container (or at least a portion of the side wall of the container detected by the M images from the camera).

[0035] In this invention, it is sufficient that a first series of images and a second series of images are acquired, each series of images representing the unfolding of the outer surface of the container in a complete 360° rotation around the axis of symmetry. However, any number N of series of images can be acquired during a first period, i.e., all are acquired while the container is rotating at the same first constant angular velocity, without any acceleration or deceleration (or acceleration and deceleration caused by existing tolerances and precision levels in the machine, unrelated to this discussion).

[0036] The number of images in a series N, and consequently the number of aggregated images N, is preferably between 2 and 10. At least two series of images are required for this invention. If N > 10, generally there is no advantage in terms of accuracy, and it is merely a waste of time and resources.

[0037] When a container rotates around its axis of symmetry at a first constant angular velocity ω, upon reaching dynamic balance, the liquid contained within rotates firmly with the container (i.e., relative motion between the container and the liquid is lost), and the free surface takes on the concave shape of a paraboloid of revolution.

[0038] The phenomenon of shape change in a rotating liquid is due to the influence of internal frictional forces acting between fluid elements and friction between the liquid and the container wall during the period from the start of rotation to the achievement of a dynamic balance state. Internal frictional forces are determined by several factors, including the viscosity of the fluid, the contact area, and the relative velocity between fluid elements.

[0039] At the start of rotation, the liquid elements in contact with the outer sidewalls of the container begin to move in relative motion. Since a state of dynamic balance has not yet been achieved, there is not enough centripetal force to keep them in a circular orbit, and the motion of the outermost elements pulls the innermost elements towards the wall due to internal friction. Because the angular velocity is constant, the relative velocities cancel each other out at the ends, and dynamic balance is achieved in a parabolic shape.

[0040] In a state of dynamic balance, each infinitesimal element of a liquid, having a mass dm = pdV (where p and dV are the density and volume of the element, respectively), traces a circular orbit and exerts the following force: a gravitational force, which is a vertically downward volume force proportional to the element's mass dm and the coefficient pdVg (where g represents the acceleration due to gravity).

number

[0041] In an inertial reference device, volume force

number

number

number

[0042] Along the vertical, that is, along the reference axis parallel to the axis of symmetry of the container when the container is placed on a stand, the volume forces and surface forces are in equilibrium, and therefore the resultant force of the surface forces exerted by the fluid on a small element of the fluid is equal to the gravitational force of the displaced volume of fluid dV, in the opposite direction, and is called the Archimedean thrust.

[0043] In the radial direction, no volume forces act, and changes in the applied pressure along this direction are centripetal forces that maintain the element dm in a circular orbit around the axis of symmetry, which is also the axis of rotation.

number

[0044] The origin of the centripetal force required for the rotation of a minute element of a liquid can be precisely determined by the pressure exerted on its tiny volume as it fluctuates along the radial direction.

[0045] When fine particles with density p'≠p and mass p'V are suspended in a liquid, the pressure exerted by the surrounding liquid is always the same, but the centripetal force may not be sufficient to maintain the particles in a circular orbit of radius r. In particular,

[0046] When the density of particles or impurities is lower than the density of the liquid, i.e., when the density of the fine particles is lower than the density of the liquid, the fine particles tend to move toward the axis of rotation. The category of fine particles with a density lower than the fluid contained in the container includes bubbles that may form in the liquid.

[0047] On the other hand, if the density of the particles is greater than the density of air, the particles tend to move away from the axis of rotation and towards the outer sidewall of the container. If the resulting force on the particles is not sufficient to counteract the viscous friction, the Archimedean thrust acting on them is not sufficient to balance the gravitational force on the particles, and the particles begin to fall towards the bottom. Otherwise, if the resulting force is greater than the viscous friction exerted by the fluid, the particles continue to move away from the axis of rotation until they reach the outer sidewall of the container. At this point, if the rotational speed is such that the gravitational force on the particles balances the static friction between the particles and the wall, the particles remain attached to the wall and continue to rotate together with the container.

[0048] As presented above, potential defects within a container may behave differently when the liquid inside rotates around an axis of symmetry. Upon reaching a state of dynamic balance, "lighter" bubbles containing air will move toward the axis of rotation, while impurities, which are likely to have a higher density, will move toward the side walls of the container, where they will stop and remain attached to the walls without moving.

[0049] In other words, the difference between impurities and bubbles lies in their behavior during the rotation step. At some point during rotation, there is no longer any relative motion between the container and the impurities (because the impurities have already adhered to the sidewalls). In a steady state, the liquid is still moving relative to the container (in fact, bubbles that do not move to the outer wall remain suspended in the liquid). The latter, due to their own inertia, move relative to the container, and this "separation" between the liquid and the sidewalls causes bubbles to move relative to the sidewalls, unlike impurities. This "separation" due to the liquid's inertia lasts for a period of time corresponding to the liquid's viscosity. Therefore, the useful time interval for this step, i.e., the step in which "heavy" impurities adhere to the wall and "light" bubbles still move, is from when the impurities adhere to the wall (when the relative velocity of the impurities to the sidewalls is almost zero) until before the liquid moves together with the wall. During the time interval when impurities are attached to the sidewalls (but before the liquid moves together with the sidewalls), image acquisition starts and ends before the liquid itself reaches the same velocity as the container. The duration depends on the size of the container and the viscosity of the liquid.

[0050] Therefore, under these conditions, the impurities maintain an orbit integral to the container, and as a result, their position on the side walls of the container remains unchanged during the rotation of the first period. Bubbles tend to move within the liquid during every rotation step, even at a constant angular velocity. In fact, since bubbles have a lower density than the liquid they are immersed in, the radial forces they experience during the rotation of the container are not sufficient to keep them in a circular orbit toward the walls of the container.

[0051] This difference in behavior can be highlighted by comparing the first and second sets of images with each other. For example, the first and second aggregated images can be compared with each other.

[0052] The two sets of images are taken as the container rotates at a first constant angular velocity. Therefore, the value of the constant angular velocity is assumed to be such that the impurities are pressed against the side walls of the container. It is also assumed that the acceleration time is sufficient to ensure that the fluid reaches a "stationary" state, i.e., a state where there is no relative rotation between the container and the impurities, i.e., sufficient for the impurities to adhere to the sides of the container.

[0053] During a first period at a first constant angular velocity vmax, in which first and second aggregated images, or more generally, N aggregated images representing only N unfoldings of the side walls of the container, are obtained, the rotational speed of the container about its own axis must be such that the impurities are pressed close to the inner wall of the container by the action of centrifugal force and maintain that position throughout the first period.

[0054] In each series of images or aggregated images, regions that may represent defects, whether impurities or bubbles, are identified. Defects are distinguishable from a simple liquid in that the pixels that identify the defect in the image have at least one characteristic that differs from the characteristics of the surrounding pixels. Generally, in aggregated images or series of images, the image pixels have very similar characteristics for the most part. This is because, in the ideal case, the majority of pixels in the aggregated image exhibit the characteristics of an “image of the liquid” in a container, just as a liquid contains no defects and can therefore be seen through the side walls of the container itself. Thus, in the first or second aggregated image (or in the first or second series of images), several pixels that generally have at least one characteristic that differs from the characteristics of the surrounding pixels or clusters are identified and can usually be clustered. A “defect region” is then defined as a region identified in the first or second aggregated image (or in the first or second series of images) that has pixels with characteristics that differ from the characteristics of the surrounding pixels. Alternatively, a defect region contains pixels with substantially uniform characteristics.

[0055] The comparable characteristics may include intensity, color, or one or more of these. For example, a defective region is a region that has a different grayscale intensity compared to the intensity of the rest (most) of the aggregated image.

[0056] Defect regions in the first and second aggregated images, or in the first and second series of images, have their positions. Regarding the position of a defect region to be determined, for example, the position of the centroid of the defect region can be considered as the position of the region. Alternatively, the position of a defect region is given by the position of its centroid. Alternatively, the position of a defect region is given by the positions of all the pixels that constitute it.

[0057] Its position is given by coordinates within the image (either M individual images or an aggregated image). As mentioned above, an image is formed by an array of pixels, each of which has its own unique coordinates. Therefore, on the first and second aggregated images, or the first and second series of images, the defect region is identified by its characteristics and location, and first and second maps of the defect region are generated.

[0058] Therefore, it is possible to compare the map of defect regions identified in the first aggregated image (or the first series of images) with the map of defect regions identified in the second aggregated image (or the second series of images). Each defect region identified in the first aggregated image has two possibilities. The first possibility is that in the first aggregated image, the first defect region is located at a first position, and in the second image, the first defect region is located at a first position that corresponds to the first position of the first defect region in the first image. A corresponding position means that the coordinates in which the defect region is located in the first aggregated image are the same as the coordinates in which the defect region is located in the second aggregated image.

[0059] The related images are generated such that the same coordinates in the first and second aggregated images correspond to the same points within the side wall of the container.

[0060] Furthermore, in order to say that the same defect region exists in both the first and second aggregated images, the two defect regions do not need to be in exactly the same location in both aggregated images; that is, the defect region does not need to be in exactly the same corresponding location in both the first and second images, but the area around that location is also considered. In other words, once the location of the defect region in the first image is identified, it is checked whether there is a defect region in the second image in the corresponding location and the surrounding area.

[0061] The area may be, for example, within a range of 10 pixels in each direction centered on the corresponding location. Preferably, the range is 10 pixels or less. More preferably, the range is 5 pixels or less. The range can also be changed and set based on the image resolution and the typical size of impurities in a given liquid. Dedicated image processing software can define an area around the defect region and quickly compare a first related image with a second related image to check whether the same defect is located at the same location in both images, or in both cases, within the surrounding area of ​​the same location.

[0062] Preferably, in an optional step of the present invention, it is further checked that substantially the same defect exists in both the first and second associated images. This can be done, for example, using a threshold operator if the pixels are grayscale. For example, the threshold is imposed on the gray (or color) level of the pixel. Dedicated software analyzes each pixel of the associated first and second images one by one, and if the gray level of a particular pixel exceeds the threshold, that pixel is considered defective. To confirm that it is the same defect, the same threshold must also be exceeded in the second image.

[0063] If three or more aggregated images are generated, for example, if N related images are generated, this comparison is performed across all aggregated images. The defect region is identified in all N aggregated images. Then, it is evaluated whether the defect region present in the first aggregated image is also present in the second aggregated image, and so on, from the third aggregated image up to the nth aggregated image. Each positional comparison between one aggregated image and the next is always performed considering the area around the location of the defect region. That is, if in image j+1 the defect is located in the area around the same location as the defect region in image j, then the same defect region is present in both image j and image j+1.

[0064] In the case of N aggregated images, the defect region does not need to be present in all N aggregated images. For example, it can be thought of as a persistence index, where the defect region is considered to be present in all N aggregated images if it exceeds a certain threshold in any case. In other words, it is sufficient for the defect region to be present "in most cases".

[0065] Therefore, by comparing the aggregated images, it can be determined which are the "persistent defect regions" or the first impurities, i.e., defect regions found at the same location in all aggregated images.

[0066] These persistent defect areas identify impurities, not bubbles.

[0067] Therefore, the method and apparatus of the present invention can determine whether impurities or bubbles are actually present in a liquid contained in a container. In fact, “persistent” defect regions, i.e., defect regions seen in both the first and second aggregated images (or the first and second series of images), are actually impurities, as can be seen, because the rotation of the container induces them to lean against the sidewall of the container and remain attached to it. Bubbles, on the other hand, which may appear as defect regions in only one of the two images, are not persistent because they are “light,” they do not adhere to the sidewall during rotation, they do not move at the same speed as the sidewall, and they have their own inertia. In other words, they are “at the mercy” of the movement of the liquid in the container.

[0068] Therefore, a simple comparison of images allows for the determination of whether or not impurities are present, distinguishing between bubbles and impurities, and as a result, an excessive number of containers will not be discarded due to bubbles being mistakenly identified as impurities.

[0069] Preferably, the camera is sensitive to electromagnetic radiation in the visible light or infrared range. These two types of radiation are most suitable for inspection because they do not involve any particular danger or safety protocols.

[0070] Preferably, the apparatus includes a light source of a predetermined electromagnetic radiation, which is positioned on the opposite side of the container from the camera. To minimize the source of errors and to acquire images under the same lighting conditions, the light source is positioned to backlight the container.

[0071] Preferably, the camera is a linear camera, and each of the first or second series of images includes a plurality of linear images acquired at a fixed angular interval. Preferably, the processing unit controls the rotating device and the camera so that there is synchronization such that each angular interval corresponds to a linear image. The aggregated image is simply a combination of linear images taken at a fixed angular interval.

[0072] Preferably, the processing unit is • Control the rotating device to stop the container from rotating. • Rotate the container again at a second angular velocity, and control the rotating device to keep the second angular velocity constant over a second period. • While rotating at a second constant angular velocity, the camera is controlled until it acquires at least a third and fourth series of images, where each of the third and fourth series of images is an image of a portion of the container, and each of the third and fourth series of images represents the portion of the container rotated 360°. • In the third and fourth series of images, the defect region is defined, and each defect region has at least one characteristic that is different from the characteristics of the adjacent region, and third and fourth maps of the defect region are generated, each map having the location and characteristics of the defect region, and the same location in the third and fourth maps identifies the same location in the container. • Compare the locations of the defect areas in the third and fourth maps. If a defect region exists at a location in the third map, and another defect region exists in the area surrounding the same location in the fourth map, then it is proven that the second impurity is present in the container or in the liquid inside the container. - Compare the locations of the first impurity and the second impurity, and if, at one location of the first impurity in the first or second map, there is no corresponding second impurity in the area surrounding the same location in the third or fourth map, then it is proven that impurities are present in the liquid placed in the container. It is further programmed in this way.

[0073] Preferably, this method is • A step to stop the rotation of the container, The steps include: rotating the container again at a second angular velocity and keeping the second angular velocity constant over a second period; The steps include: acquiring at least a third and a fourth series of images while rotating at a second constant angular velocity, wherein each of the third and fourth series of images is an image of a portion of the container, and each of the third and fourth series of images represents the portion of the container rotated 360°; The steps include: identifying defective regions in a third and fourth series of images, each defective region having at least one characteristic different from the characteristics of adjacent regions, and generating third and fourth maps of the defective regions, each map comprising the location and characteristics of the defective region, wherein identical locations in the third and fourth maps identify the same location within the container; - A step of comparing the locations of the defect areas in the third and fourth maps, - If a defect region is located in a third map and a defect region is located in the area surrounding the same location in a fourth map, the step is to prove that a second impurity is present in the container or in the liquid contained in the container. - A step to verify that impurities are present in the liquid placed in the container if, at one of the locations of the first impurities in the first or second map, there is no corresponding second impurity in the area surrounding the same location in the third or fourth map. Includes.

[0074] To distinguish whether the first impurity is an impurity present inside or outside the liquid, the present invention provides an optional step. In fact, if the impurity is present inside the liquid, the container is likely to have to be discarded. However, if the impurity is present externally, the container may be retained. External impurities may also include defects in the container itself, such as cracks or bubbles in the material forming the sidewalls of the container. To distinguish between the two types of impurities, the container is stopped while rotating at a first constant angular velocity for a first period. After being stopped for a specific period, the container is rotated again. The container is moved at a second angular velocity which is kept constant over a second period. The second rotational velocity about the axis of symmetry of the container itself must be such that the internal impurity is pressed close to the inner wall of the container by the action of centrifugal force and maintains its position relative to the sidewall over the second period. Preferably, the first angular velocity is equal to the second angular velocity. Preferably, the first period has a duration equal to that of the second period. During this second period, the camera is activated, for example via a processing unit, and as a result acquires images of portions of the container, wherever they are located within its field of view or wherever selected portions of the container are located, at angular intervals, preferably at constant angular intervals. Furthermore, the camera remains active, i.e., continues to acquire images for at least first and second full rotations of the container about the container's axis of symmetry. The second period must be long enough to allow the camera to acquire images of portions of the container that have radial symmetry with respect to the first and second full rotations of the container over 360°.

[0075] During each first or second rotation, the camera acquires M images. These images are acquired in the same manner as in the first period.

[0076] The M images acquired in the first rotation and the M images acquired in the second rotation form the third and fourth sets of images. These images are preferably associated with each other. This association is preferably performed by a processing unit. In the association, an aggregated image is generated for each set of M images. Thus, the association of the third set of M images generates the third aggregated image, and the association of the fourth set of images generates the fourth aggregated image. This association is performed in the same way as for the first and second aggregated images.

[0077] Preferably, if N aggregated images are acquired in the first period, then N aggregated images are also acquired in the second period.

[0078] Each pixel in the third and fourth aggregated images uniquely corresponds to a single point on the side wall of the container (excluding the line at the end of the image, as is known). Therefore, each pixel coordinate in the third and fourth aggregated images corresponds to an exact point on the side wall of the container. The same coordinates in the first, second, third, or fourth aggregated images correspond to the same point on the side wall of the container.

[0079] In the third and fourth aggregated images, regions that may represent defects, whether impurities or bubbles, are identified. Region identification is performed in the same manner as described for region identification in the first and second aggregated images. Subsequently, “defect regions” are identified in the third and fourth aggregated images.

[0080] In the third and fourth aggregated images, the defective regions have their locations. Therefore, on the third and fourth aggregated images, the defective regions are identified by their characteristics and locations, and third and fourth maps of the defective regions are formed.

[0081] Therefore, just as with the first and second maps of the defect regions, it is possible to compare the map of the defect regions identified in the third aggregated image with the map of the defect regions identified in the fourth aggregated image.

[0082] Therefore, by comparing the third and fourth aggregated images, it can be determined which is the "persistent defect region" or second impurity, i.e., the defect region found at the same location in all aggregated images. This comparison is performed in exactly the same manner as with the first and second aggregated images.

[0083] Therefore, multiple "first persistent defect regions" or first impurities identified by analyzing aggregated images detected during rotation in the first period, and multiple "second persistent defect regions" or second impurities identified by analyzing aggregated images detected during rotation in the second period, are available. When the container underwent deceleration, stopping, and new acceleration between the first and second periods, the impurities present in the liquid within the container that were firmly "adhered" to the sidewalls during rotation at the first angular velocity were separated therefrom. When the container returns to its original motion, the impurities are pressed against the sidewalls again, but in a different position than where they were during rotation at the first angular velocity. Therefore, by comparing the position of the first impurities with the position of the second impurities, if there is no equivalent for each first impurity in the same position defining the second impurity, this means that the impurity has moved and therefore is an impurity within the container. On the other hand, if the first impurity corresponds to the second impurity at the same location, this impurity does not undergo displacement when the liquid accelerates or decelerates, and therefore means that it is located on the outer surface of the container or is a defect in the side wall itself.

[0084] Preferably, this method includes the following steps, or a processing unit: • Determine the difference map of the position of the first impurity, • Determine the difference map of the position of the second impurity, - Compare the positions of the first and second impurities. If, at one position of the first impurity in the first difference map, there is no corresponding second impurity in the area surrounding the same position in the second difference map, then it is proven that impurities are present in the liquid placed in the container. It is further programmed in this way.

[0085] A difference map is simply a map containing persistent defect regions and their locations. The first difference map contains a first impurity having characteristics and locations, as obtained from the analysis of aggregated images acquired during a first period. The second difference map contains a second impurity having characteristics and locations, as obtained from the analysis of aggregated images acquired during a second period.

[0086] Preferably, the density of the liquid in the container is less than 2000 centipoise. The liquid in the container must be sufficiently fluid to allow for relative rotation with respect to the side walls of the container.

[0087] Preferably, this method is • A step to stop the rotation of the container, The step of rotating the container again at a second angular velocity and In between, • The third step of keeping the container shut down for a period of time. Includes.

[0088] After the first period, the rotation is stopped to ensure that impurities are "separated" from the inside of the container, and the rotation is only restarted after the "stopped" period.

[0089] Preferably, the first or second angular velocity is 200 rpm to 10,000 rpm. This angular velocity is sufficient to achieve the necessary dynamic balance in commonly used containers and to move common impurities such as hair, glass, metal, and rubber towards the container walls.

[0090] Preferably, this method includes the step of backlighting the container using the predetermined electromagnetic radiation. This minimizes errors due to changes in camera illumination.

[0091] Preferably, this method is The step of acquiring a series of N images (2 ≤ N ≤ 10) while rotating at a first or second angular velocity (vmax1, vmax2), wherein each series of images represents a portion of the container rotated 360°.

[0092] If 2 ≤ N ≤ 10, good accuracy can be obtained when detecting impurities without excessively slowing down the inspection.

[0093] Preferably, the step of identifying defect regions in a first and second series of images includes the step of analyzing the pixels that make up the first and second series of images, and the step of identifying clusters of pixels having different characteristics from pixels adjacent to the cluster as defect regions. Impurities typically form regions of a small number of pixels in an image.

[0094] Further features and advantages of the present invention will be described in more detail in the following description of preferred and non-limiting examples of the embodiments. [Brief explanation of the drawing]

[0095] [Figure 1] A schematic top view of an apparatus for inspecting a container containing liquid, as inspected according to one embodiment of the present invention, is shown. [Figure 2] This shows a side view of a container holding the liquid to be inspected according to the present invention. [Figure 3a] A schematic top view of the inspection apparatus according to the present invention in the subsequent inspection step is shown. [Figure 3b] A schematic top view of the inspection apparatus according to the present invention in the subsequent inspection step is shown. [Figure 3c] A schematic top view of the inspection apparatus according to the present invention in the subsequent inspection step is shown. [Figure 4] The graph shows the rotational speed of the container as a function of time according to the steps of the present invention. [Figure 5] This document shows multiple images and their analysis based on the steps of the present invention. [Modes for carrying out the invention]

[0096] Referring to the above figures, the apparatus for inspecting the container 8 according to the present invention is generally shown as 4.

[0097] Container 8 contains a liquid that is transparent or partially transparent to a predetermined electromagnetic radiation, for example, in the visible range, and is transparent or at least partially transparent to the same electromagnetic radiation, at least in part thereof. The liquid is preferably for medical use.

[0098] Container 8 is at least partially or partially transparent to electromagnetic radiation so that the liquid placed inside can be seen from the outside.

[0099] In the preferred example shown in the figure, the container 8 comprises a side wall 20 that is transparent or partially transparent to electromagnetic radiation, a bottom wall 26, and a cap 25, as can be clearly seen in Figure 2. The cap 25 is, for example, not transparent to electromagnetic radiation. The liquid is placed in the portion of the container 8 bounded by the side wall 20 and the bottom wall 26. The side wall and the bottom wall are preferably formed as a single unit. The side wall 20 and the bottom wall are preferably made of glass. The side wall 20 defines an inner surface 23 that comes into contact with the liquid and an outer surface 22 (as can be clearly seen in Figure 1).

[0100] Furthermore, container 8 has a radial axis of symmetry X with respect to at least a portion of it. Container 8 in the illustrated example has cylindrical symmetry, or in other words, it is a solid of rotation around the axis of symmetry X.

[0101] The inspection device 4 includes a rotating device 5 for the container 8, which is adapted to support the container 8 and includes a motor means 6 ( schematically shown as a rectangle in Figure 1) for rotating the container 8 around an axis of vertical rotation that coincides with the cylindrical axis of symmetry X of the container 8.

[0102] The apparatus 4 further includes a camera 12 positioned to acquire an image of a portion 16 of the side wall 20 of the container 8 in its field of view, for example, in the form of pixels. Preferably, the portion 16 is such that its extension along axis X includes the bottom wall 26 and terminates at a height above the height defined by the liquid level in the container 8. The camera 12 includes a dedicated sensor that is sensitive to electromagnetic radiation through which the liquid and the container are at least partially transparent.

[0103] Since the container 8 is equipped with side walls 20 that are transparent or partially transparent to electromagnetic radiation that the camera is sensitive to, the camera 12 can acquire not only an image of the side walls 20 of the container 8, but also an image of its contents (i.e., liquid).

[0104] The object of the present invention is to detect the presence of defects in a container and to distinguish whether such defects (if any) are actually internal impurities 14 or bubbles 15 in the liquid.

[0105] Furthermore, as shown in Figure 2, in addition to the bubbles 15 and internal impurities 14 present in the liquid, the container 8 may also exhibit external impurities / defects 17 on the outer surface 22 of the side wall 20 of the container 8. External defects 17 may be, for example, damage to the container (e.g., cracks) or external impurities present on the outer surface 22. If the impurities are on the inner surface 23, they will come into direct contact with the liquid placed inside the container 8 and therefore should preferably be discarded, but if the defects are on the outer surface 22, the container can still be used.

[0106] The apparatus 4 preferably includes an illumination device 24 positioned on the opposite side of the container 8 from the camera 12, thereby backlighting the container 8 on the opposite side of the camera 12 using electromagnetic radiation that the camera is sensitive to. The illumination device 24 is, for example, a planar illumination panel.

[0107] The apparatus 4 further comprises a processing unit 32 operably connected to the rotating device 5 and the camera 12.

[0108] According to the method of the present invention, the processing unit 32 is programmed as follows: The rotating device 5 is controlled to rotate the container 8 around the rotational symmetry axis X until a constant and predetermined angular velocity is achieved.

[0109] Figure 4 shows a graph of the angular velocity of the container as a function of time. As can be seen in Figure 4, the processing unit 32 controls the rotating device 5 to accelerate the container 8 with acceleration acc1 until it reaches a predetermined angular velocity vmax1 which is kept constant for a first period t1. During this period t1, the container 8 makes N complete rotations over 360° around the rotation axis X. In each of these N rotations, the processing unit 32 controls the activation of the camera 12 to acquire M images of the portion 16 at a predetermined regular angular interval 18Δθ of a number of M, as shown in Figures 3a to 3c.

[0110] Camera 12 is, for example, a linear camera, with a constant angular interval dθ1=dθ2=…=dθ M A linear image 40 is acquired at Δθ, and for each 360° rotation around the container's axis X, a continuous unfolding of the container's side 20 is generated, as detailed below. The camera 12 uses a rotary encoder connected to a motor means 6 that rotates the container to acquire M images in sequence, i.e., one image 40 for every Δθ°.

[0111] The width of the angular interval in which the linear camera 12 scans the unfolding of the container 8 is determined by the resolution obtained. The value of the angular interval may be equal to, for example, 0.0012 radians.

[0112] By combining all M linear images 40 acquired by camera 12, an aggregated image 50 is obtained, as shown in Figure 3c. Thus, the aggregated image 50 is an unfolded representation of the entire side wall 20 of the cylindrical container 8.

[0113] The process shown in Figures 3a-3c is repeated for all N rotations that occur in the first period t1. Thus, as can be seen in Figure 4, a first series of aggregated images 50 are generated. The first series consists of S1, S2…S N It is called, and here, each S i (i=1, ..., N) is the aggregated image 50. N is equal to, for example, 5.

[0114] As shown again in Figure 4, during the acceleration step acc1, the camera can still acquire images, but these are not considered in subsequent processing. The time it takes for the container 8 to reach a first predetermined angular velocity vmax1 from a stationary state is, for example, 500 milliseconds, and is chosen so that at the start of period t1, the liquid is no longer in relative motion with respect to the outer wall of the container 8. Thus, at the start of the first period, the experimental conditions are such that impurities 14 present inside the container 8 and having a "high" density come into contact with and are pressed toward the inner surface 22 of the side wall 20, and therefore do not move substantially during rotation, while bubbles 15 move within the liquid. In other words, unfold S1, S2…S N During the first period of a constant angular velocity vmax1 in which the impurity 14 is obtained, the rotational speed of the container 8 about its own axis of symmetry X must be such that the impurity 14 is pressed against the inner surface 22 of the container 8 by the action of centrifugal force and maintains its position.

[0115] Next, the first series S1, S2...S NEach aggregated image 50 is analyzed as shown in Figure 5. Pixels in each aggregated image 50 are examined, and defective regions 60 that have one or more characteristics different from the pixels in the surrounding region are identified. In fact, most pixels in image 50 are considered to have uniform characteristics by representing a uniform image of the liquid contained in container 8. Therefore, there may be relatively few regions where pixels exhibit characteristics different from the characteristics of most pixels in aggregated image 50. These “different” regions are highlighted in each of the first series of aggregated images 50. The “different” characteristics for identifying the regions 60 may be, for example, color intensity. These regions in each aggregated image are called defective regions 60. As seen in Figure 5, this operation is performed on each of the first series of aggregated images 50. Thus, a “map of defective regions” is generated on each of the series of aggregated images, and each defective region 60 has a precise location.

[0116] The defect region may represent an external defect 17 on the outer surface 22 of the side wall 20, a bubble 15 in the liquid, or an impurity 14 in the container.

[0117] To distinguish the defect regions 60 and determine which are actually internal impurities and not bubbles, then the first series S1, S2…S N A comparison of all aggregated images 50 is performed using the processing unit 32. For each defective region 60 present in the first aggregated image S1, it is checked whether the defective region is located at the same position or in at least one surrounding area in the second aggregated image S2. In other words, if a defective region is found in the first aggregated image at a particular position, it is checked whether the defective region is also found in the series of second aggregated images at the same corresponding position or in an area around that corresponding position.

[0118] In the case of N aggregated images, a search is performed in all N aggregated images to determine whether the defective region is located at the same position or in an area surrounding the same position.

[0119] For example, for each defective region 60, the position of its centroid is calculated, and a search is performed in all N images to determine if there is a defective region in a 10-pixel area around the corresponding position of the centroid.

[0120] In the case of N aggregated images, in order to determine if a defect region exists in all of them, the defect region 60 is a first sequence S1, S2...S N The defect does not need to be exactly repeated in all 50 aggregated images. For example, a persistence index is used. If a defect region is in the first aggregated image, it is searched to see if the defect region is in the same location (at least the surrounding area) in the second aggregated image. If this is true, a specific score is assigned. If this is not true, a considerably lower score is assigned. The same comparison and scoring is then performed on a third set of images, and so on. Then all scores are summed up, and if this sum exceeds a threshold, the defect region 60 is considered to be present in all of the first set of aggregated images 50.

[0121] "Persistent" defect regions, i.e., defect regions found in all of the first series of aggregated images, are considered impurities 70. In fact, if a defect region is persistent, it exhibits "persistence" of the defect at a specific location even during rotation of the container. These impurities 70 may be both internal impurities 14 and external defects / impurities 17, but they may not be bubbles 15, as bubbles change location between aggregated images. The first difference map 80 of impurities 70 is then generated by comparing all defect regions 60 in all of the first series of aggregated images 50.

[0122] If it is desirable to distinguish between internal impurities 14 and external impurities 17, the above process is repeated over an additional period of time.

[0123] Referring again to Figure 4, the processing unit 32 controls the rotating device 5 to decelerate the container 8 at a deceleration rate dec1. Preferably, the coefficients of dec1 and acc1 are the same. The container 8 is then stopped, for example, 500 milliseconds. Preferably, the container 8 remains stopped for a third time tstop. For example, time tstop is equal to 50 milliseconds.

[0124] Next, the container 8 is rotated again until it reaches a second predetermined angular velocity vmax2. Preferably, vmax1 = vmax2. As can be seen in Figure 4, the processing unit 32 controls the rotating device 5 to accelerate the container 8 at an acceleration acc2 until it reaches vmax2, which is kept constant over a second period t2. Preferably, acc1 = acc2. During this period t2, the container 8 makes N complete rotations over 360° around the rotation axis X. The number of rotations N performed in the second period t2 is preferably equal to the number of rotations N performed in the first period t1. In each of these N rotations in the second period t2, the processing unit 32 controls the activation of the camera 12 to acquire M images of the portion 16 for a predetermined number of regular angular intervals 18Δθ, as shown in Figures 3a to 3c.

[0125] As described above for the first period, a second series of aggregated images 50 are subsequently acquired, and as shown in Figure 5, a map of the defect region 60 is formed in each of the second series of aggregated images 50. Thus, similar to the first series, the second series S1, S2…S N A second difference map 90 is formed in which only the "persistent" defect regions 100 in all of the aggregated images 50 are preserved.

[0126] Next, a comparison is made between the first difference map 80 and the second difference map 90.

[0127] As the container 8 decelerates and then accelerates, the impurities 14 present inside the container 8 are "separated" from the inner surface 22 and, upon reaching a second constant angular velocity, are pressed against the inner surface 22 again, but almost certainly in a position different from the position assumed during the first period.

[0128] On the other hand, external impurities or defects 17 of the container always remain in the same position and therefore maintain the same position in any difference map.

[0129] Therefore, by comparing the locations of the persistent defect region 70 in the first difference map 80 and the persistent defect region 100 in the second difference map 90, the first persistent defect region in the first difference map corresponds to the defect region in the second difference map in the same location or surrounding area, and the type of defect is an external impurity or defect on the outer surface of the side wall 20 of the container. This is the case, for example, of the "crack" shown by 17, which is seen in the same location in the first and second difference maps 80 and 90 in Figure 5. On the other hand, if the second persistent defect region in the first difference map is no longer seen in the same location in the second difference map, but the defect region still exists in a different location, then the impurity 14 is present in the container, and therefore the container must be discarded. This case can always be seen in Figure 5, where a persistent "moving" defect region is seen in the first and second difference maps (shown by 14).

[0130] Therefore, when comparing the two difference maps, only the pixels whose intensity values ​​have changed are highlighted. This is due solely to the presence of impurities occurring at different points in the first difference map 80 compared to the second difference image 90.

[0131] This comparison between two difference maps can be performed, for example, as a simple calculation of the difference between pixel intensities in the two difference maps.

Claims

1. An apparatus (4) for inspecting a container (8) containing a liquid that is at least partially transparent to a predetermined electromagnetic radiation, wherein the container (8) comprises at least one part having radial symmetry about an axis of symmetry (X), and the apparatus (4) - A rotating device (5) adapted to rotate the container (8) about the axis of symmetry (X), - A camera (12) positioned so as to be within the field of view when the container (8) is placed, and the camera (12) is sensitive to the predetermined electromagnetic radiation, - A processing unit (32) adapted to control the rotating device (5) and the camera (12), - The container (8) is rotated at a first angular velocity (vmax 1), and the rotating device (5) is controlled to maintain the first angular velocity constant over a first period (t1). - The camera (12) is controlled to acquire at least a first and second series of images while rotating at the first constant angular velocity (vmax 1), wherein each of the first or second series of images (40) is an image of a portion (16) of the container (8), and each series of images represents the portion of the container rotated 360°. - In the first and second series of images (S1, S2) (40), a defect region (60) is identified, each defect region has at least one characteristic different from the characteristics of adjacent regions, first and second maps (S1, S2) of the defect region are generated, each map (S1, S2) comprises the location and characteristics of the defect region, and the same location in the first map and the second map identifies the same location in the container (8), - Compare the positions of the defect regions in the first and second maps (S1, S2), - If a certain defect region (60) is located in the first map (S1), and a certain defect region is located in the area surrounding the same location in the second map (S2), then the presence of a first impurity (70) distinct from bubbles in the liquid is demonstrated, including impurities (17) present on the outer surface (22) of the side wall (20) of the container (8) and impurities (14) present in the liquid contained in the container (8). The processing unit (32) is programmed to do so and A device (4) comprising:

2. The apparatus (4) according to claim 1, wherein the camera (12) is sensitive to electromagnetic radiation in the range of visible light or infrared light.

3. The apparatus (4) according to claim 1, comprising the aforementioned predetermined electromagnetic radiation light source (24), wherein the light source (24) is positioned on the opposite side of the container (8) from the camera (12) so as to backlight the container (8).

4. The apparatus (4) according to claim 1, wherein the camera (12) is a linear camera, and each of the first or second series of images (M) includes a plurality of linear images acquired at a constant angular interval (Δθ).

5. The processing unit (32) is - Control the rotating device (5) to stop the container (8), - Rotate the container (8) again at a second angular velocity (vmax2), and control the rotating device (5) to keep the second angular velocity constant over a second period (t2). - While the rotation is performed at the second constant angular velocity, the camera (12) is controlled until it acquires at least a third and a fourth series of images, each of the third and fourth series of images (M) being an image (M) of a portion (16) of the container (8), and each of the third and fourth series of images representing the portion of the container (8) rotated 360°, - In the third and fourth series of images, a defective region (60) is identified, each defective region having at least one characteristic different from the characteristics of an adjacent region, third and fourth maps of the defective region are generated, each map comprising the location and characteristics of the defective region, and the same location in the third map and the fourth map identifies the same location in the container (8), - Compare the positions of the defect regions (60) in the third and fourth maps, - If a defect region is located in the third map and a defect region is located in the area surrounding the same location in the fourth map, the presence of a second impurity distinct from bubbles in the liquid is demonstrated, including impurities (17) present on the outer surface (22) of the side wall (20) of the container (8) and impurities (14) present in the liquid contained in the container (8). - The positions of the first impurity (70) and the second impurity are compared, and if the second impurity in the area surrounding the same position in the third or fourth map does not match the position of the first impurity (70) in the first or second map, it is proven that the impurity (14) is present in the liquid placed in the container (8). The apparatus (4) according to claim 1, further programmed as described above.

6. - A step of providing a container (8) that is at least partially transparent to a predetermined electromagnetic radiation and contains a liquid that is at least partially transparent to the predetermined electromagnetic radiation, wherein the container (8) comprises at least one part having radial symmetry about an axis of symmetry (X), - The steps of rotating the container (8) at a first angular velocity (vmax 1) and keeping the first angular velocity constant for a first period of time, - A step of acquiring at least a first and a second series of images during rotation at the first constant angular velocity, wherein each of the first and second series of images (M) is an image of a portion (16) of the container, and each series of images represents the portion of the container (8) rotated 360°. - A step of identifying a defective region (60) in the first and second series of images, wherein each defective region (60) has at least one characteristic different from the characteristics of an adjacent region, and generating first and second maps (S1, S2) of the defective region (60), wherein each map includes the location and characteristics of the defective region, and the same location in the first map and the second map identifies the same location in the container (8), - A step of comparing the positions of the defect regions (60) in the first and second maps (S1, S2), - If a certain defect region (60) is located in the first map (S1) and a certain defect region is located in the area surrounding the same location in the second map (S2), the step of proving the presence of a first impurity (70) which is distinct from bubbles in the liquid, including impurities (17) present on the outer surface (22) of the side wall (20) of the container (8) and impurities (14) present in the liquid placed in the container (8). A method for inspecting a container (8), including the following.

7. - A step of stopping the rotation of the container (8), - The steps of rotating the container (8) again at a second angular velocity (vmax2) and keeping the second angular velocity constant over a second period (t2), - A step of acquiring at least a third and a fourth series of images during the rotation at the second constant angular velocity (vmax2), wherein each of the third and fourth series of images (M) is an image of a portion (16) of the container (8), and each of the third and fourth series of images represents the portion of the container (8) rotated 360°. - A step of identifying a defective region (60) in the third and fourth series of images, wherein each defective region has at least one characteristic different from the characteristics of an adjacent region, and generating third and fourth maps of the defective region, each map comprising the location and characteristics of the defective region, wherein the same location in the third map and the fourth map identifies the same location in the container, - A step of comparing the positions of the defect regions (60) in the third and fourth maps, - If a defect region is located in the third map and the defect region is located in the area surrounding the same location in the fourth map, the step of proving the presence of a second impurity, which is distinct from bubbles in the liquid, and which includes impurities (17) present on the outer surface (22) of the side wall (20) of the container (8) and impurities (14) present in the liquid contained in the container (8). - A step of comparing the positions of the first impurity (70) and the second impurity, and if the second impurity in the area surrounding the same position in the third or fourth map does not match the position of the first impurity (70) in the first or second map, then proving that the impurity (14) is present in the liquid placed in the container (8). The method according to claim 6, including the method described in claim 6.

8. The method according to claim 6, wherein the viscosity of the liquid placed in the container (8) is less than 2000 centipoise.

9. - A step of stopping the rotation of the container (8), - A step of rotating the container (8) again at a second angular velocity (vmax2) In between, - A step of keeping the container (8) stopped for a third period of time. The method according to claim 7, including the method described in claim 7.

10. The method according to claim 7, wherein the first or second angular velocity (vmax1, vmax2) is 200 rpm to 10000 rpm.

11. The method according to claim 7, wherein the first constant angular velocity is equal to the second constant angular velocity.

12. The method according to claim 6, further comprising the step of illuminating the container (8) with a backlight using the predetermined electromagnetic radiation.

13. The method according to claim 7, comprising the step of acquiring a series of N images (2 ≤ N ≤ 10) during the rotation at the first or second angular velocity (vmax1, vmax2), wherein each series of images represents the portion of the container (8) rotated 360°.

14. The method according to claim 6, wherein the step of identifying a defective region (60) in the first and second series of images includes the step of analyzing pixels that make up the first and second series of images (M), and the step of identifying a cluster of pixels having different characteristics from the pixels adjacent to the cluster as a defective region.

Images