Optical inspection method and system

Abstract

The present invention relates to a method of optically inspecting bottles comprising producing one or more digital representations by optically sensing one of the bottles, and automatically analyzing the one or more digital representations for identifying anomalies, wherein the optically sensing comprises optically sensing an external appearance of the one of the bottles, preferably while the uptake volume is empty, such that at least one of the analyzed digital representations characterizes the external appearance, and wherein the optically sensing comprises optically sensing an internal appearance at least of a bottom of the one of the bottles through the opening while the uptake volume is empty such that at least one of the analyzed digital representations characterizes the internal appearance of the bottom. Further, the present invention relates to a system and a computer program as well as a computer-readable medium for performing the method.

Description

[0001] Optical Inspection Method and System

[0002] The present invention generally relates to a method of optically inspecting bottles, a system, a computer program and a computer-readable storage medium.

[0003] In particular, the present invention relates to optically inspecting vials, i.e. bottles of preferably small volumes like 500 ml or less, typically used for storing vaccines or other medicaments. Bottles or vials can be inspected with or without a closing device like a stopper or a septum.

[0004] Optical inspection of bottles and vials is well known in the art as a manual task during production, in particular in the pharma industry. In the known process, bottles like vials are filled with a preferably pharmaceutical substance and closed with a septum or different stopper. Afterwards, while the vials are continuously passing a human inspector, visual inspection is performed by means of human eyes of the inspector. Once the inspector identifies an anomaly that is regarded unacceptable, like a damage of the vial or foreign particles in the vial, the respective vial is manually discarded.

[0005] The known optical inspection suffers from potential inaccuracy depending on human competence and fitness which sometimes demands for second inspection. Further, the known approach results in significant reject costs since not only the bottle, but also the content, i.e. the pharmaceutical substance, has to be discarded upon identification of an unacceptable anomaly.

[0006] Thus, it is an objective problem of the present invention to improve quality assurance and to reduce the reject costs.

[0007] This is achieved by a method according to claim 1 , by a system according to claim 16, by a computer program according to claim 17, and by a computer-readable storage medium according to claim 18. Advantageous embodiments are subject to the dependent claims.

[0008] One aspect of the present invention relates to a method of (comprising a step of) optically inspecting bottles, in particular vials, the method comprising for each of several bottles under inspection: producing one or more digital representations by optically sensing one of the bottles, and automatically analyzing the one or more digital representations for identifying anomalies, wherein the optically sensing comprises optically sensing an external appearance of the one of the bottles, in particular while an uptake volume is empty, such that at least one of the analyzed digital representations characterizes the external appearance, and wherein the optically sensing comprises optically sensing an internal appearance, preferably of the surface, at least of a bottom of the one of the bottles through an opening of the one of the bottles while the uptake volume is empty such that at least one of the analyzed digital representations characterizes the internal appearance of the bottom.

[0009] The analysis is performed such that anomalies like defects or particles are searched for in the digital representations. Generally, experienced humans are very good in identifying anomalies and in classifying them to be acceptable or unacceptable compared to known automatic identification by means of digital representations. However, according to the present invention, a synergistic combination of analyzing both digital representations of the external appearance of the bottle as well as of at least of the bottom area of the internal volume of the bottle, in the following also referred to as internal appearance, turned out particularly reliable and efficient even in comparison with traditional manual inspection.

[0010] Analyzing digital representations preferably covers to find / identify anomalies by means of searching for visually recognizable features in the digital representation, like a particular pattern, shape, color, color gradient, brightness, brightness gradient, size, arrangement, texture, contrast, direction, spatiality, movement and the like. However, the features preferably are not limited to those indicated.

[0011] Identifying anomalies particularly preferably comprises recognizing and classifying anomalies. Classifying can cover determining a level of anomaly, e.g., as to whether a scratch is serious. An anomaly preferably is identified if the anomaly is relevant, exceeds a threshold or, generally speaking, is unacceptable.

[0012] Unless required by the individual context, it is not necessary to identify acceptable anomalies. That is, it is possible to implement a one-step approach of identifying merely unacceptable anomalies. If no acceptable anomalies are searched for or identified, the analysis preferably comprises both identification / finding anomalies and classification / qualification as unacceptable in a single step while it is possible to ignore acceptable anomalies. Consequently, finding and afterwards classifying / qualifying anomalies, i.e., analyzing in two steps is preferred, but also possible in a joint process, in particular if Al based recognition is involved. For the sake of conciseness, the one-step approach and the two-step approach in the following are covered in that the term “anomaly” without qualification can be replaced with “unacceptable anomaly” unless this replacement contradicts the individual context. Sometimes, anomalies are qualified as “(unacceptable) anomalies” to emphasize that the “unacceptable” is preferred but not mandatory.

[0013] The analysis of the digital representations preferably covers automatic search for patterns potentially originating from anomalies, i.e., for appearances or structures in the digital representations that are unexpected or potentially originate from defects, foreign substances like particles and the like that might affect the usability of the bottle.

[0014] Two or more different digital representations of the same bottle or a part thereof can be analyzed together in order to identify anomalies, e.g., by identifying a structure that is represented differently in the digital representations when there is an anomaly rather than expected. For example, an anomaly like a chip or crack might result in a discrepancy between different digital representations of the same region, and an anomaly thus can be identified by finding such discrepancies. This can be supported by Al like discussed further in the following.

[0015] In one aspect of the present invention, at least two digital representations originating from different ones of optically sensing an internal appearance, optically sensing an external appearance, optically sensing an opening appearance, or optically sensing a closure appearance of the / each one of the bottles are analyzed together. Preferably, a potential anomaly is identified or verified by analyzing digital representations of the same area of the bottle from different origins I perspectives.

[0016] For example, a digital representation of the surface, at least of a bottom of the one of the bottles, through an opening of the one of the bottles while the uptake volume is empty, and a digital representation of an external appearance of the bottom, e.g., through a side wall of said bottle, are analyzed together. Analyzing different digital representations of the same bottle together can cover identifying discrepancies or similarities between the digital representations. In particular, the digital representations can be merged, superposed or compared for finding, preferably optical, discrepancies or similarities between the digital representations. Identified discrepancies or similarities between different digital representations then can be qualified as anomalies or not.

[0017] Further, preferably, analyzing the digital representations covers classification of identified anomalies. The analysis preferably comprises differentiating anomalies into acceptable and unacceptable anomalies, confirming and opposing usability, respectively.

[0018] An example for acceptable anomalies is a production variation confirmed to fall within an acceptable tolerance, while foreign substances like a loose glass splinter inside the bottle, serious scratches or entrapped air are examples for unacceptable anomalies.

[0019] In case an anomaly is identified and classified unacceptable, the respective bottle can be automatically discarded.

[0020] Examining the empty bottle before filling advantageously enables withdrawing / removing empty bottles, and preferably avoids filling them, thus without the need to discard the bottles including content. This saves resources and costs.

[0021] Taking into account both the external appearance and the internal appearance, preferably of the internal surface at least of the bottom of the bottle, enables identifying anomalies like foreign substances, e.g. particles like loose splinters, inside the bottle which can hardly be reliably found by an analysis of the external appearance only. Analyzing the external appearance of the bottle, on the other hand, turned out particularly advantageous to identify anomalies related to integrity of the bottle forming material.

[0022] Particularly advantageously, the external and internal appearances or respective analysis results can be combined to verify as to whether an anomaly is present, acceptable or unacceptable, in particular when the anomaly can be identified in both representations. Consequently, the present invention surprisingly results in improved quality assurance due to the particularly reliable anomaly identification and classification and in reduction of reject costs due to early examination and discarding still empty bottles having unacceptable anomalies.

[0023] A further aspect of the present invention, which can be realized independently as well, relates to a packaging method comprising the steps of the method of automatically optically inspecting bottles for storing medical products, in particular vials, according to the present invention as described above and further comprising monitoring a pharmaceutical substance for anomalies on its way to be filled into the one of the bottles.

[0024] The method of optically inspecting bottles according to the present invention turned out particularly advantageous in the context of packaging. In particular, the packaging method relates to a packaging of a pharmaceutical substance being liquid or dry. The pharmaceutical substance can be inserted into the bottle, which then can be closed and optionally be covered by an outer packaging. Discarding bottles having unacceptable anomalies according to the method of the present invention enables increasing efficiency of the packaging process as the rejection rate late in the packaging process can be kept low.

[0025] A further aspect of the present invention, which can be realized independently as well, relates to a packaging method comprising the steps of the method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to one or more aspects of the present invention as described above, the method comprising filling the one of the bottles only if no anomaly of the one of the bottles is identified when analyzing the digital representation's characterizing the internal and / or external appearance of the bottle under inspection, and to prevent filling the one of the bottles if analyzing the digital representation's results in identifying one or more anomalies, preferably when the anomalies are classified unacceptable.

[0026] It turned out particularly advantageous to discard bottles having unacceptable anomalies already and just before the filling step of the packaging process in order to avoid loss of content, in particular pharmaceutical substance, when the already filled bottle has to be discarded. A further aspect of the present invention, which can be realized independently as well, relates to a method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to one or more aspects of the present invention as described above, the method comprising for each of several bottles under inspection: producing one or more digital representations by optically sensing one of the bottles and automatically analyzing the one or more digital representations for identifying anomalies, wherein the automatically analyzing is performed using at least one machine learning model which is trained using digital representations of several different sample bottles, some of which have anomalies and the remaining of which have no anomalies represented in the respective digital representation, and using pre-known anomaly properties qualifying each of the sample bottles concerning anomalies.

[0027] Applying at least one machine learning model to the digital representation(s) has turned out particularly efficient and reliable to find (unacceptable) anomalies and / or to classify anomalies as acceptable or unacceptable. In this regard, the at least one machine learning model is preferably trained with multiple representations of bottles having various anomalies and having no anomalies, together with information about the presence or absence of (unacceptable) anomalies and / or classification as acceptable or unacceptable.

[0028] A “machine learning model” in the sense of the present invention preferably comprises and / or uses a machine learning algorithm and one or more (training) datasets. In other words, a machine learning model is preferably the output of a machine learning algorithm run on one or more (training) datasets. In particular, a machine learning model represents what was learned by a machine learning algorithm. An algorithm is in particular a procedure that is run on one or more (training) datasets to create the machine learning model. The algorithm may in particular be an artificial neural network.

[0029] A “training dataset” in the sense of the present invention is preferably a set of data that is used to train the machine learning model and / or machine learning algorithm.

[0030] An “artificial neural network” in the sense of the present invention is preferably a computational learning system that uses a network of functions to understand and translate a data input of one form into a desired output, usually in another form. A neural network is composed of at least two, preferably three or more layers. In particular, an artificial neural network has an input layer, an output layer and one or more hidden layers, i.e. layers between the input layer and the output layer. Each layer has one or more units called “neurons”. The concept of artificial neural networks is inspired by the brain and the way a brain learns.

[0031] A further aspect of the present invention, which can be realized independently as well, relates to a method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to one or more aspects of the present invention as described above or hereinafter, the method comprising for each of several bottles under inspection: producing at least two different digital representations by differently optically sensing (the same) one of the bottles, preferably from different perspectives (e.g., outer and inner appearance), and automatically analyzing the digital representations together for identifying anomalies, wherein the automatically analyzing is performed using at least one machine learning model which is trained using respective digital representations of several different sample bottles, some of which have anomalies and the remaining of which have no anomalies represented in the respective digital representation, and using pre-known anomaly properties qualifying each of the sample bottles concerning anomalies. For training the machine-learning model, thus, for each bottle the different digital representations are input. The output, for example, can be set to “anomaly identified” or “no anomaly identified” or corresponding information.

[0032] A further aspect of the present invention, which can be realized independently as well, relates to a method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to the present invention as described above or hereinafter, the method comprising for each of several bottles under inspection: inputting different digital representations produced by differently optically sensing (the same) one of the bottles, preferably from different perspectives (e.g., of outer and inner appearance), and procuding an output with the machinelearning model, the output preferably being anomaly properties qualifying each of the sample bottles concerning anomalies, like “anomaly identified” or “no anomaly identified” or corresponding information. Thus, the machine-learning model can be used to particularly efficiently search for and reliably identify anomalies, in particular of vials, by combined processing the analysis of digital representations of different optical sensing processes (different pictures from different perspectives). A further aspect of the present invention, which can be realized independently as well, relates to a method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to the present invention as described above, the method comprising for each of several bottles under inspection: producing one or more digital representations by optically sensing one of the bottles, and automatically analyzing the one or more digital representations for identifying anomalies, wherein the optically sensing comprises optically sensing at least one, preferably an external, surface of the one of the bottles up to a depth of less than 100 micrometer, preferably less than 50 micrometers, in particular 20 micrometers or less, such that at least one of the analyzed digital representations characterizes the surface merely up to said depth.

[0033] Sensing by scanning merely the limited depth from the surface surprisingly turned out to enable more efficient and reliable identification of anomalies, in particular defects. The limited depth can be obtained - preferably depending on the material from which the bottle is formed - by choosing a suitable device, procedure, radiation wavelength applied to the bottle and / or a suitable perspective or angle, in particular such that a penetration depth of the radiation is essentially limited accordingly, or, alternatively or additionally, such that reflected or transmitted components of the radiation are essentially limited accordingly.

[0034] A further aspect of the present invention, which can be realized independently as well, relates to a method of automatically optically inspecting bottles for storing medical products, in particular vials, preferably according to the present invention as described above, the method comprising for each of several bottles under inspection: producing one or more digital representations by optically sensing one of the bottles, and automatically analyzing the one or more digital representations for identifying anomalies, wherein the optically sensing comprises optically sensing an internal appearance of the one of the bottles using an endoscopic sensing device being inserted into the one of the bottles such that at least one of the analyzed digital representations characterizes the internal appearance.

[0035] Thus, it is preferred to generate digital representations of the internal appearance of the bottle, preferably at least of the area of the bottom of the bottle. These digital representations can be analyzed for identifying anomalies. It turned out particularly efficient and effective to generate such digital representation by means of an endoscopic sensing device inserted into the uptake volume of the bottle. On the one hand, quality limiting effects, e.g., due to scattered light, can be reduced or avoided. Further, this enables a favorable perspective in which optionally the side walls can be included in the digital representation, and later analyzed if desired. At least transitions between the bottom and the side walls of the bottle can be sensed particularly reliably enabling a more precise anomaly identification.

[0036] An endoscopic sensing device preferably can be inserted, e.g., into the uptake volume at least partially such that at least part of the sensor for producing the digital representation (which preferably is responsible for collecting information like radiation, e.g. light) is or can be placed within the uptake volume.

[0037] A further aspect of the present invention, which can be realized independently as well, relates to monitoring content by means of a particle detection device for detecting and / or qualifying particles in the content before entering the bottles.

[0038] The particle detection device comprises a line for supplying content to a filling station and a window for monitoring the content, wherein the particle detection device comprises at least one adapter for adapting the line to the window such that content flowing through the line passes the window in a flow direction.

[0039] The window preferably has a geometry that differs from a geometry of the line. In particular, the window realizes a passage for the content having a flat cross-section.

[0040] The particle detection device preferably detects particles in the content using a particle sensing device. Alternatively or additionally, the particle detection device qualifies particles to be anomalies based on a composition of the detected particles detected using a composition sensing device.

[0041] A further aspect of the present invention, which can be realized independently as well, relates to a system for optical inspection of bottles for storing medical products, in particular vials, comprising: one or more optical sensors for optically sensing one of the bottles under inspection for producing one or more digital representations by optically sensing the one of the bottles, and an analysis module for automatically analyzing the one or more digital representations for identifying anomalies, wherein the system is configured to execute one of the methods according to the present invention.

[0042] The system according to the present invention implements the method according to the present invention and, thus, enables the technical effects and advantages indicated above and in the following.

[0043] The system comprises means for performing the method, preferably at least one sensing device for sensing the bottle and detecting the digital representation, and an analysis device or analysis module being configured to analyze the digital representations in order to identify anomalies, and preferably to classify them as unacceptable or acceptable.

[0044] An analysis module in the sense of the present invention can receive or retrieve the digital representation and can analyze the digital representation for finding and / or identifying anomalies and / or for qualification / categorization / classification of anomalies and / or for determining a suitability for further use for packaging or distribution.

[0045] A further aspect of the present invention, which can be realized independently as well, relates to a computer program comprising instructions to cause the system of the present invention to execute the steps of one of the methods according to the present invention.

[0046] The computer program is preferably configured to control the system, in particular the sensing device to sense the digital representation(s) of the bottle and / or the analysis device or analysis module to analyze the result, i.e. the digital representation, for identifying the anomalies.

[0047] A computer program preferably is a product comprising code to be executed on a processor or being suitable to be converted into code for performing a method like controlling sensors and / or analyzing digital representations.

[0048] Finally, a further aspect of the present invention, which can be realized independently as well, relates to a computer-readable medium having stored thereon the computer program according to the present invention. The technical effects and features discussed in context with the computer program apply to the computer-readable medium as well.

[0049] Bottles in the sense of the present invention preferably have an uptake volume for storing medical products, wherein the uptake volume may be delimited / bounded by an internal surface of the respective bottle forming the internal appearance at least partially or essentially, the uptake volume being accessible via an opening of the bottle. The internal surface is formed by a side wall and a bottom. At the outside, the bottle has an external surface forming the external appearance at least partially or essentially. The internal appearance or external appearance may be affected by a view through the surface depending on the bottle’s transparency. The opening can be closed by a closing device like a stopper or septum, preferable a rubber stopper or septum.

[0050] A bottle in the sense of the present invention preferably has side walls preferably being essentially cylindrical and the bottom being attached to and bordering in a tight manner the side walls in order to form a container. Preferably, the bottle or container has its opening opposite to the bottom and / or connected to the side walls by means of a bottle neck.

[0051] An uptake volume in the sense of the present invention preferably is a space surrounded by walls or other barriers, thus being an inner space or being able to hold content. The uptake volume can be accessed via an opening. However, it is preferred that with closed opening(s), e.g. if a stopper or septum closes off the opening(s), the uptake volume is sealed such that the content like liquid (pharmaceutical preparation) contained in the uptake volume is prevented from exiting / escaping.

[0052] An internal surface in the sense of the present invention preferably is the surface bordering the uptake volume.

[0053] Vials in the sense of the present invention are bottles of preferably small volumes typically used for storing vaccines or other medicaments. The vaccines or other medicaments may be for injection purposes. Such bottles or vials usually have a crimp neck and can be closed with or without a closing device like a stopper or a septum. The closing device may comprise a crimp cap securing the stopper or septum at the crimp neck. A septum preferably is a stopper for closing off the opening. A septum is preferable perforable or penetrable multiple times by means of an injection needle without losing its sealing effect once the needle is removed.

[0054] Although the present invention is particularly favorable in the context of vials (before and / or after filling and closing), the present invention can be applied to different bottles as well. Further, the present invention is particularly efficient in the context of bottles made from glass or different transparent or at least translucent material, but the present invention is not limited thereto.

[0055] Optically inspecting in the sense of the present invention preferably is an optical examination for identifying anomalies. Optical inspection can be carried out by producing at least one digital representation of at least part of the bottle and an analysis of this digital representation being suitable for finding and, preferably, for classification of anomalies as acceptable or unacceptable. The optical inspection of the present invention preferably is an automated process involving at least one sensing process, in particular imaging or scanning process, and producing the digital representation thereby.

[0056] Optical or visual inspection generally preferably covers determining the absence or presence of visible particles within parenteral products. Parenteral products are those that are administered by injection or infusion, such as vaccines, blood products, or intravenous fluids. Visual inspection is performed by trained human operators using their naked eye while optical inspection in the sense of the present invention preferably is performed by automated machines using cameras and sensors.

[0057] Visual or optical inspection is an essential operation in the pharmaceutical industry to ensure the safety and quality of the drug products and their containers. It can be used to detect defects such as cracks, leaks, discoloration, foreign matter, or incorrect labeling. Visual or optical inspection can follow standardized procedures and guidelines that specify the illumination, ambient conditions, inspection time, and acceptance criteria.

[0058] Anomalies in the sense of the present invention are properties that vary or differ from expected results. Anomalies in particular are or cover contaminations like particles or other foreign substances adhering to the bottle or being contained in the bottle, and / or potential defects like damages or enclosures of the bottle, and / or of the bottle shape.

[0059] Anomalies in the sense of the present invention preferably are or comprise defects, foreign substances like particles, loose glass splinters and the like, that might affect the usability of the bottle, or appearances, patterns or structures in a digital representation being unexpected or potentially originating from defects, foreign substances like particles, loose glass splinters and the like, that might affect the usability of the bottle.

[0060] When inspecting the filled and closed bottle, anomalies can be identified at the closing device or the content alternatively or additionally to identifying them at the (empty) bottle itself.

[0061] Identifying anomalies preferably covers applying an imaging process on at least a part of the bottle. Further preferably, identifying covers analyzing the digital representation, which results from the imaging process, for anomalies.

[0062] In a preferred embodiment, identifying anomalies comprises classification of the digital representations, of parts thereof or of anomalies found therein based on evaluation bases for finding and ranking anomalies. Ranking anomalies can either result in qualification of an anomaly as acceptable or unacceptable, but alternatively or additionally can result in a quantity qualification which in a further step can result in categorization as acceptable or unacceptable.

[0063] The invention can comprise different classification processes, preferably at least one for finding anomalies within the digital representations and another one for finding out whether an anomaly is acceptable (bottle can be used) or unacceptable (bottle is to be discarded). However, alternatively or additionally, the identification process can be used for identifying merely unacceptable anomalies.

[0064] An unacceptable anomaly in the sense of the present invention preferably is a deviation from an expected result potentially causing harm, potentially affecting shelf life, potentially affecting mechanical stability and / or potentially affecting physical or chemical integrity of the content inside or to be filled into the bottle or vial. An acceptable anomaly in the sense of the present invention preferably is a deviation from an expected result which, however, does not affect the function of the bottle or vial or its content such that it can be ignored.

[0065] A digital representation in the sense of the present invention preferably is digital data representing the physical properties of the appearance or shape. Particularly preferably, it can be a picture or scan or generally speaking a result from an imaging technology or imaging technique.

[0066] Optically sensing in the sense of the present invention preferably is a method of technically recognizing an appearance, in particular a surface and / or shape, of an object. Particularly preferably, optically sensing is or covers an imaging technique like taking one or multiple picture(s) or video(s), confocal imaging, confocal microscopy or shadow measurement, and / or a scanning technique like laser scanning, and / or other optical imaging and / or scanning techniques. Different techniques for optical sensing can and preferably are combined, in particular used sequentially.

[0067] Automatically analyzing in the sense of the present invention preferably needs or covers processing the digital representation, i.e. the results originating from the optically sensing process, preferably an image or scan. The processing preferably comprises at least one analysis being suitable for identification, preferably covering finding and / or classification (assigning to a particular class or category, in particular ranking as regards (un)acceptability for further use) of anomalies. For example, the digital representation can be analyzed for identification of patterns known to be related to a particular anomaly like a foreign substance adhering to the bottle, lying on the bottom inside the bottle, like a splinter and the like.

[0068] Particularly preferably, the automatically analyzing or processing of the results originating from the optically sensing process, i.e. of the digital representations, is or covers an Al (Artificial Intelligence) based method and / or using a machine learning model. This can be achieved by an artificial intelligence-based I machine learning model and / or matrix, which preferably is or has been trained with a number of bottles being free of anomalies and having anomalies, the latter preferably being ranked as acceptable and unacceptable anomalies such that it enables identification and / or classification of anomalies. Using an Al based automated analysis has surprisingly turned out particularly efficient and, thus, is preferred. An appearance in the sense of the present invention preferably is what a contactless imaging or scanning technology / technique is able to recognize and provide or store as a digital representation. In other words, an appearance in the sense of the present invention preferably is what an optical sensor can recognize from an object. An external appearance preferably is an appearance when sensing the external surface. An internal appearance preferably is an appearance when sensing the internal surface.

[0069] In case the material used enables the appearance to be influenced by the material beyond the respective surface, the appearance might comprise optical components or information from beyond the surface accordingly.

[0070] Preferably, the bottle is transparent. Anyway, the external appearance is the result when a sensing direction is from outside while the internal appearance remains the result when a sensing direction is from inside.

[0071] Sensing an internal appearance in the sense of the present invention preferably is performed either by sensing through the opening or by inserting the sensor through the opening, both resulting in gathering the digital representation directly or merely through the uptake volume.

[0072] Sensing an external appearance preferably comprises sensing the external surface which can be influenced from beyond. However, sensing an external appearance does preferably not comprise sensing (from the outside) directly through the opening as the result thereof represents an internal appearance in the sense of the present invention, even if a sensor is outside but has a perspective to produce the digital representation from the internal surface.

[0073] A surface of a bottom in the sense of the present invention preferably is (a section of) a surface opposite to the opening which preferably can be sensed in an opening direction through the opening or from outside. The surfaces of the bottom can border the surfaces of side walls.

[0074] A search for patterns preferably comprises digitally processing the digital representation(s) for content / patterns being similar to content / patterns known to originate from anomalies. In particular, search for patterns can comprise performing similarity assessment of the whole or some parts, preferably stepwise, of the digital representation(s), based on the digital representation(s) or aggregations thereof to one or more references being known from or assigned to an anomaly.

[0075] Discarding a bottle in the sense of the present invention can comprise, or can be limited to marking the bottle, e.g. by assigning said bottle a marker physically, electronically or virtually, enabling the bottle to be selected, to be withdrawn, and / or to be removed from the process for use or disposal. Alternatively or additionally, discarding the bottle can directly be a disposal.

[0076] Packaging in the sense of the present invention preferably comprises covering or sealing or bottling up the content to be distributed, preferably pharmaceutical substance, appropriate for commercialization. On the one hand, packaging can comprise filling the content into the bottle. Alternatively or additionally, packaging comprises closing the bottle and / or adding an outer packaging for distribution purposes, optionally along with additional distribution or use information. Nevertheless, packaging in the sense of the present invention can be limited to filling and sealing the bottle or vial.

[0077] A pharmaceutical substance in the sense of the present invention preferably is substance being able to be used for treating illnesses. The pharmaceutical substance can be a liquid or a dry substance for forming a liquid or a suspension like a lyophilizate for injection purposes. However, although the present invention turned out to be particularly advantageous in this regard, the invention is not necessarily limited to the background of or does not necessarily form part of a process for packaging and distribution of pharmaceutical substances.

[0078] Scanning in the sense of the present invention preferably is or comprises a technique to produce a digital representation of an appearance recognizing the appearance section by section, for example by means of a moving laser beam, e.g., line by line via the surface.

[0079] A scanning depth in the sense of the present invention preferably describes an optical penetration extent essentially perpendicular to a surface area into the material forming the appearance like the bottle or vial, or its wall, bottom or other part thereof. Optical sensors according to the present invention can be radiation-based imaging and / or scanning sensors. Optical sensors can be passive (without radiation / light emission, e.g. for receiving external light) or active (with radiation / light emission in order to receive reflected or transmitted light). Optical sensors in the sense of the present invention preferably can be one or more of a camera, laser scanning device, shadow imaging device, confocal microscopy device, and the like.

[0080] Further aspects of the present invention can be gathered from the figures and the following discussion of preferred embodiments referring to the figures.

[0081] In the figures:

[0082] Fig. 1 depicts a proposed system for optical inspection;

[0083] Fig. 2 depicts exemplary bottles / vials having various anomalies; and

[0084] Fig. 3 depicts a particle detection device.

[0085] Fig. 1 depicts a system 1 for optical inspection of bottles 2. The bottles 2 preferably are for storing medical products as content 3 in an uptake volume 4.

[0086] Fig. 2 depicts some examples bottles 2 with possible anomalies 14. Of course, many other anomalies 14 are known by the person skilled in the art and can be identified by the method and system 1 of the present invention.

[0087] Referring to the bottle 2 according to Fig. 2A and 2B, the bottles 2 preferably have a side wall 5 and bottom 6 forming the uptake volume 4. Usually opposite to the bottom 6 an opening 7 allows filling content 3 into and withdrawing content 3 from the uptake volume 4.

[0088] The bottles 2 particularly preferably are vials, but the invention can be applied to other bottles 2 as well. Again, referring to the example depicted in Fig. 2A, vials can be bottles 2 having a crimp neck 8. A crimp neck 8 preferably provides an undercut positive-locking to a beading of the crimp cap 11.

[0089] The present invention surprisingly has turned out particularly advantageous in context of processing vials. However, the invention can be applied to other types of bottles 2, while in the following the invention primarily is described with a bottle having the opening 7 formed by a "crimp neck" 8. Thus, the term “crimp neck” can be replaced with the term “bottle neck” unless indicated to the contrary. Either the crimp neck 8 or any bottle neck surrounds the opening 7 of the bottle 2 and delimits it while a face end can form a contact area for a closing device 9.

[0090] The bottle 2 or vial can be configured to be closed with the closing device 9, preferably being or comprising a stopper 10 and / or septum which is optionally secured by a crimp cap 11 that can form part of the closing device 9, e.g., as depicted in Fig. 2C to 2F. However, the inspection is applied at least partially to the empty, open bottle 2 or vial without closing device 9 as discussed in the following.

[0091] The bottle 2 has an external appearance and an internal appearance. The external appearance preferably originates and / or results from a view at an external surface 12. The internal appearance preferably originates and / or results from a view at an internal surface 13 of the bottle 2.

[0092] The bottle 2 can long-term store content 3. The bottle 2 is tight for the content 3 and can be sealed with the closing device 9 whereby the bottle 2 or uptake volume 4 is completely enclosed such that no content 3 can escape due to a change in orientation of the bottle 2. The content 3 preferably is a pharmaceutical liquid, solution, or suspension, but can also be a dry substance like a powder, a lyophilizate or the like. However, the majority of aspects of the present invention are related to an empty bottle 2 or at least do not demand for content 3 to be inside the uptake volume 4.

[0093] Moreover, the present invention is particular advantageous in pharmaceutical related cases, e.g., due to the high-quality standards in this field, but can be applied to and be advantageous in different fields of technology, i.e. for bottles 2 with or for different content 3 or use, as well.

[0094] The system 1 and the corresponding method is for optically inspecting each of several bottles 2 under inspection for identifying anomalies 14. Anomalies 14 primarily are deviations from a bottle 2 according to a specification, e.g., deviations due to contamination, defects, or production tolerances beyond specified requirements. Anomalies 14 can be acceptable, e.g., if a defect is classified uncritical or a dust particle on the external surface 12 might be ignored in particular cases. An anomaly 14 can be a production variation resulting in an uncritical optical effect. However, usually anomalies 14 are unacceptable, in particular when affecting the content 3 or the uptake volume 4 cannot be ruled out. For example, dust particles or splinters inside the uptake volume 4 are usually classified as unacceptable anomalies 14.

[0095] According to the present invention, digital representations 15 of at least one of the bottles 2 are produced by optically sensing the one of the bottles 2.

[0096] A digital representation 15 usually is data representing at least one appearance of the bottle 2. Digital representations 15 typically are either picture or scan data. Nevertheless, other digital representations 15 like a model representing the bottle 2 or generally results of an aggregation process preferably on picture or scan data are possible as well. The digital representations 15 do not need to represent the whole bottle 2. However, it is appreciated that digital representations 15 represent the bottle 2 to a sufficient extend to enable qualification of the bottle 2 as acceptable or unacceptable, or to find anomalies 14 with a sufficient probability.

[0097] The proposed system 1 can comprise one, but preferably comprises two or more different sensing devices 16 to 20 for generating digital representations 15.

[0098] An external appearance sensing device 16 can be provided to produce a digital representation 15 of the external appearance, in particular the or covering the external surface 12, at least partially. In the example depicted in Fig. 1 , the external appearance sensing device 16 is a camera. Thus, the digital representation 15 of the external appearance can be one or more photos and / or videos. When the bottle 2 is transparent, the external appearance can comprise information regarding the content 3.

[0099] In Fig. 2A, a cross section of a bottle 2 is depicted comprising a particle in the uptake volume 4. This particle potentially can be found by analyzing the digital representation 15 produced with the external appearance sensing device 16. However, this cannot be guaranteed.

[0100] An internal appearance sensing device 17 can be provided to produce a digital representation 15 of the internal appearance, in particular the or covering the internal surface 13. In the example depicted in Fig. 1 , the internal appearance sensing device 17 is a camera. It can be endoscopic for insertion through the opening 7 or have a telephoto lens for making a digital representation 15 from outside the bottle 2, but with view through the opening 7. In the sense of the present invention, both alternatives, when the sensing device 17 is inserted or has a view through the opening 7, are regarded as “sensing through the opening 7”. Again, the digital representation 15 of the internal appearance can be one or more photos and / or videos.

[0101] Figure 2B is an example of a digital representation 15 generated with internal appearance sensing device 17 through opening 7. In Fig. 1 , the internal appearance sensing device 17 is inserted into the uptake volume 4 and has a viewing angle at least covering the bottom 6. Alternatively, the internal appearance sensing device 17 can be placed outside the bottle 2, i.e. not inserted into the uptake volume 4, but with view through the opening 7. The digital representation 15 resulting therefrom is depicted in Fig. 2B. Here, the anomaly 14 in form of a particle laying on the bottom 6 can be found by means of the analysis according to the present invention. Actually, the particle found, i.e. the anomaly 14, can be verified by means of the digital representation 15 produced by means of the external appearance sensing device 16 provided that the particle or at least a shadow or artifact corresponding thereto can be identified in this digital representation 15 of the external appearance sensing device 16. Such check preferably forms part of the proposed analysis.

[0102] Preferably, one or more particle detection devices 28 is / are provided, being configured to monitor the content 3 in advance of dispensing the content 3 into the bottle 2. With other words, the substance which later is to be inserted in the uptake volume 4 and then is the content 3 of the bottle 2, is or can be monitored. Particularly preferably, the particle detection device / -s 28 is / are attached to or forms part of a filling station 21. The particle detection device / -s 28 preferably is / are configured to generate a digital representation 15 like a photo or a video of the content 3, i.e. the substance to be dispensed into the bottles 2, in order to identify foreign substances like particles as anomalies 14. If an anomaly 14 is identified with the particle detection device 28 the bottle 2 in which it is inserted can be associated with the anomaly, e.g., for direct or later disposal.

[0103] The bottle 2 can be filled with the content 3 by placing the content 3 in the uptake volume 4, preferably in a filling station 21 for dispensing the content 3 into bottles 2. Filling can be effected, preferably automatically, by means of content 3 being dispensed into the uptake volume 4 of the bottle 2, e.g., with a pipette. In the example shown, the filling station 21 comprises multiple dispensing elements like pipettes for simultaneously filling multiple bottles 2. However, filling can be effected differently as well.

[0104] After filling is completed, the bottle 2 can be closed which can be achieved by placing the closing device 9, for example realized by or comprising the stopper 10, in particular septum, being shown in Fig. 2A, inside and sealing the opening 7. Optionally, the stopper 10 of the closing device 9 can be secured afterwards with the crimp cap 11 of the closing device 9 shown in Fig. 2A. Preferably, the crimp cap 11 can be placed over the stopper 10 and the crimp neck 8 and can be crimped such that the septum or stopper 10 of the closing device 9 is pressed to and secured at the crimp neck 8 of the bottle 2. A bottle closing station C can be provided and form part of the system 1 being configured to close the bottle 2 by means of performing (one or more) of said steps.

[0105] The opening 7 of the bottle 2 preferably is optically inspected (optically checked for anomalies 14) after completion of filling and in the progress of closing the bottle 2 with the bottle closing station C, in particular at least one time before and one time after closing. Particularly preferably, the bottle 2 is optically inspected both directly after filling and directly after closing the bottle 2 with the closing device 9, after closing further preferably directly after closing the bottle 2 with the stopper 10, and / or directly after applying the crimp cap 11.

[0106] One or more appearance sensing devices 18A, 18B, 18C can be provided for optically inspecting the opening 7 area, in particular the bottle neck (in particular the rim of the bottle opening 7) I crimp neck 8 and / or the closing device 9, after filling of the bottle 2 being completed, i.e. , before, during and / or after closing the bottle 2.

[0107] The appearance sensing device(s) 18A, 18B, 18C can each be configured to generate a digital representation 15 like a photo or a video of the opening 7 area with and / or without closing device 9, stopper 10 and / or crimp cap 11. For the optical inspection, the digital representation 15 is generated and analyzed for identifying anomalies 14. Regarding the analysis of the digital representation 15 and handling initiated upon identification of anomalies 14, reference is made additionally to the discussion in context of external appearance sensing device 16 or internal appearance sensing device 17, as it can be the same mutatis mutandis. The unclosed opening 7 can be inspected with an opening appearance sensing device 18A, in particular inspecting the bottle neck I crimp neck 8 regarding adhesions or defects like chips and the like which can, e.g., originate from collision of the opening 7 with parts of filling station 21.

[0108] The opening 7 closed with the stopper 10, but preferably without crimp cap 11 , can be optically inspected with a fist closure appearance sensing device 18B, in particular optically inspecting the bottle neck I crimp neck 8 and / or the stopper 10 regarding adhesions or defects like chips and the like and / or a relative position of the stopper

[0109] 10 to the opening 7, i.e., correct position of the stopper 10 for closing the opening 7 or for lyophilization which is discussed in detail below. An incorrect position of the stopper 10 can, e.g., originate from a placing error when closing the opening 7 with the stopper 10. Alternatively, the opening 7 closed with another closing device 9 can be inspected accordingly.

[0110] The opening 7 closed with the stopper 10 and applied with the crimp cap 11 can be inspected with a second closure appearance sensing device 18C, in particular inspecting the bottle neck I crimp neck 8 and / or the stopper 10 and / or the crimp cap

[0111] 11 regarding adhesions or defects like bumps, ridges and the like and / or a correct position and shape of the crimp cap 11 over the stopper 10 and the crimp neck 8. Here, an anomaly 14 can originate, e.g., from an equipment for applying the crimp cap 11 which might unintendedly cause an anomaly 14 during putting a blank of the crimp cap 11 over the stopper 10 und the crimp neck 8 and / or during beading the blank of the crimp cap 11 to encompass I engage both the stopper 10 und the crimp neck 8.

[0112] The system 1 preferably comprises a transporting means T for transporting the bottles 2 along a filling appliance 26 covering at least the filling station 21 and preferably also a sterilization equipment S for (thermally) sterilizing the bottles 2 and / or a bottle closing station C for closing the bottles 2, and related inspection equipment (one or more of the external appearance sensing device 16, the internal appearance sensing device 17, the opening appearance sensing device 18A, the first closure appearance sensing device 18B, and / or the second closure appearance sensing device 18C). The transporting means T can comprise one or more conveyers like a belt or is formed thereby. However, alternatively or additionally, other concepts for the bottle 2 transport can realize or be comprised by the transporting means T.

[0113] The transporting means T preferably provides an exclusively fully automatically transport from sterilization, via filling, preferably up to closing the bottles 2. The transporting can be essentially continuous or uninterrupted. The proposed regime and equipment for inspection enables production and product safety regardless of the fully automatically transporting of the bottles 2, thus allowing for avoiding potential delays and errors which came along with manual handling which is avoided.

[0114] The transporting means T can extend and transport bottles 2 into or through the sterilization equipment S, which can be an oven for thermal sterilization.

[0115] A fully automatic transport in the sense of the present invention preferably is an automated transport which does not demand for any operator interaction. The bottles 2 may be automatically conveyed and / or may also be automatically stopped for a while during the transport, e.g., for filling. The automatic transport preferably is a transport realized completely automatically in a continuous bottle 2 processing avoiding bottles 2 to exit or enter distinct appliances.

[0116] Using other words, the automatic transport may realize a continuous bottle 2 processing along the filling appliance 26, preferably covering at least the sterilization equipment S, the filling station 21 and bottle closing station C, avoiding bottles 2 to exit or enter distinct appliances thereof, which accordingly are connected directly by means of the automatic transporting means T for the bottles 2, such that any demand of an operator intervention like withdrawing bottles 2 from one and supplying them to another appliance for processing bottles 2 is avoided.

[0117] In one aspect, which can be realized independently from the other aspects of the present invention, the system 1 is placed in a (common and / or single) clean room or the continuous bottle 2 processing according to the present invention is realized in a clean room. Thus, the complete process of inspecting the bottles 2 during the filling process as described herein preferably is realized in a clean room without bottles 2 exiting the clean room during or for transition between stations of the system 1. Particularly preferably, at least the external appearance sensing device 16 and the internal appearance sensing device 17 and, preferably, one or more of the filling station 21 , the particle detection device 28, the opening appearance sensing device 18A, the first closure appearance sensing device 18B, and the second closure appearance sensing device 18C are placed in the same clean room. This enables the bottles 2 to be filled under supervision without exiting the clean room which improves the reliability and efficiency.

[0118] In the method according to the present invention, the steps preferably are performed uninterruptedly, directly following one another, without intermediate storage, and / or without manual handling or manual handling requirement. In particular, one or more of the steps:

[0119] • sterilization,

[0120] • making a digital outer appearance representation of the bottle 2,

[0121] • making a digital inner appearance representation of the bottle 2,

[0122] • making a digital representation of the content 3 before being dispensed into the bottle,

[0123] • filling (dispensing of content 3 into the bottle 2), and / or

[0124] • making a digital representation of the opening appearance and / or of the closure of the bottle 2 (with the first closure appearance sensing device 18B and / or the second closure appearance sensing device 18C) are followed one after the other fully automatically, without any manual interaction, or without any manual interaction requirement, and further preferably within the same clean room.

[0125] In a preferred embodiment, before filling the bottle 2, the bottle 2 is conveyed from the sterilization equipment S to the filling station 21. At the filling station 21 , the bottle 2 can be filled, e.g., by automatically placing the bottle 2 at a predefined location and dispensing the content 3 in the uptake volume 4, e.g., with a pipette, nozzle, needle or different dispenser. After filling is completed, the bottle 2 is conveyed further to the bottle closing station C and be closed there. The inspection steps according to this invention further form part of the process. The whole process from sterilization to the bottle 2 being closed preferably is fully automated without demanding any operator interaction. By means of one or more of the sensing devices 16 to 20, the uptake volume can be checked or verified. While the external appearance sensing device 16 and / or the internal appearance sensing device 17 can be used for checking or verifying the total available uptake volume, one or more of the appearance sensing devices 18A, 18B, 18C can be used for checking or verifying the volume of content 3 in the uptake volume 4.

[0126] In one aspect of the present invention, it is ensured that the bottle 2 is (thermally) sterilized and then directly inspected automatically by means of internal appearance sensing device 16 and external appearance sensing device 17 while the bottle 2 is completely automatically (continuously I uninterruptedly) conveyed, preferably by means of the same transporting means T, in particular realized as conveyer. The bottle 2 may then be conveyed further to the filling station 21 and filled there.

[0127] In one aspect of the present invention, the content 4 is lyophilized in the bottle 2. For lyophilization of the content 4, the closure device 9 or the stopper 10 preferably is applied to bottle 2 such that the opening 7 is covered, but merely to an extend still enabling gas, in particular humidity, to escape. In particular, the closure device 9 or the stopper 10 is merely applied loosely to the opening 7.

[0128] During the lyophilization process, the bottle 2 is subjected to special climatic conditions (temperature, humidity) such that the content 4, which is liquid in an initial state, i.e., directly after dispensing into the uptake volume 4 of the bottle 2, is dried. The lyophilization can comprise one or more steps or variations of climatic conditions the bottle 2 or its content 4 is subjected to.

[0129] Finally, the closure device 9 or the stopper 10 is moved in a position closing the opening 7. Closing the opening 7 preferably means closing tight even for gas or humidity.

[0130] The covering of the opening 7, the lyophilization, and / or the closure of the opening 7 can be monitored, e.g., by means of the opening appearance sensing device 18A or the first closure appearance sensing device 18B, e.g., for correct position for the closure device 9 or the stopper 10, for correct lyophilization progress, and / or for correct position for the closure device 9 or the stopper 10 to close the opening 7. Hence, if the closure device 9 or the stopper 10 is in a predefined position, in which the lyophilization process is allowed to function, the position of the closure device 9 or the stopper 10 is indicative to the successful evacuation of humidity. Hence the stopper position is a quality indicator. After the lyophilization process the closure device 9 or the stopper 10 is pressed into the opening 7 for closure.

[0131] Alternatively or additionally, during the opening 7 being covered enabling lyophilization, during lyophilization in progress, and / or during the opening 7 being finally closed, the bottle 2 or its opening 7 region can be inspected (for anomalies).

[0132] A closed bottle sensing device 19 can be provided to produce a digital representation 15 of the external appearance when the bottle 2 is closed with the closing device 9. The digital representation 15 preferably is or covers the external surface 12, at least partially, in particular at least the closing device 9, i.e., preferably the stopper 10 or septum and / or the crimp cap 11. In the example depicted in Fig. 1 , the closed bottle sensing device 19 is a camera, e.g., like the external appearance sensing device 16. The external appearance sensing device 16 can be reused and form the closed bottle sensing device 19 or vice versa. The digital representation 15 of the external appearance can be one or more photos and / or videos.

[0133] Alternatively or additionally, with the closed bottle sensing device 19 the filling level of content 3 inside the uptake volume 4 of the bottle 2 can be controlled or verified after filling and closing the bottle 2 is completed.

[0134] It can be identified as anomaly 14 if the filling level does not comply with a respective requirement.

[0135] A dimension sensing device 20 can be provided, being configured to produce a digital representation 15 of dimensions and / or geometrical parameters representing the geometry of the bottle 2. The dimension sensing device 20 can be or comprise a scanner, in particular a laser scanner. The dimension sensing device 20 can be or comprise a profilometer, in particular a laser profilometer.

[0136] The system 1 can comprise a control unit 24 which can be configured to receive or retrieve from one or more of the sensing devices 16 to 20 the digital representations 15 of the bottle 2 which can be analyzed for identifying anomalies 14 by an analysis module 25. In Fig. 1 , the analysis module 25 forms part of the control unit 24. However, it can be realized separately as well. Alternatively or additionally, the control unit 24 can be configured to control one or more of the sensing devices 16 to 20 for generating digital representations 15, respectively, and / or to control the filling station 21 based on the outcome of the inspection, i.e. based on whether one or more anomalies 14 have been identified (as being unacceptable).

[0137] The proposed method comprises or the system 1 preferably is configured to automatically analyze the digital representations 15 for identifying anomalies 14. For example, a photo or video or scan of the bottle 2 can be the digital representation 15 which can be analyzed with a process being suitable to find anomalies 14 and preferably to classify them, e.g., as unacceptable.

[0138] In particular, the digital representation 15 originating from a sensing device 16 to 20 can be analyzed by means of the analysis module 25 for identifying an anomaly 14 and comparing it with a corresponding requirement in order to find out whether the anomaly 14 is acceptable. If not, the control unit 24 can initiate marking the bottle 2 as to be discarded or initiate another discarding process.

[0139] According to one aspect of the present invention, the optically sensing (process) comprises optically sensing an external appearance of the one of the bottles 2, preferably while the uptake volume 4 is empty, and preferably the opening 7 is open / unclosed, such that at least one of the analyzed digital representations 15 characterizes the external appearance of the empty, open bottle 2. The external appearance sensing device 16 can be used for this task.

[0140] Further, the optically sensing (process) preferably comprises optically sensing an internal appearance at least of the bottom 6 of the one of the bottles 2 through the opening 7 while the uptake volume 4 is empty such that at least one of the analyzed digital representations 15 characterizes the internal appearance of the bottom 6. The internal appearance sensing device 17 can be used for this task.

[0141] Sensing and analyzing digital representations 15 of both the external and internal appearance or the system 1 being configured accordingly has turned out particularly efficient in discovering and reliably classifying anomalies 14. This enables a more efficient and reliable automated inspection process. The digital representations 15 of the external appearance or sensed with the external appearance sensing device 16 and the internal appearance or sensed with the internal appearance sensing device 17 preferably are combined (processed together in combination, e.g., comparing, merging, or joined processing with a machinelearning model) as this turned out particularly resistant against incorrect marking. Particularly in case the bottle 2 is from glass or other essentially transparent material, the digital representation 15 of the external appearance enables verification of anomalies 14 identified by means of the digital representation 15 of the internal appearance, or vice versa. For example, a defect in a side wall 5 can be identified in one of the digital representations 15 and this makes an optical effect in the other one of the digital representations 15, by identification of which the anomaly 14 or its classification can be verified in a synergistic manner. In addition, the digital representations 15 can be and preferably are analyzed independently from another.

[0142] Based on identified anomalies 14, the bottles 2 preferably are marked with a marking 22. The marking 22 can be conducted for unacceptable anomalies 14. Alternatively, or additionally, a marking 22 can be conducted when finding no or merely acceptable anomalies 14. The marking 22 can be physical, e.g. a paint-based marking 22, or non-physical, in particular virtual, e.g., by continuously monitoring or assigning the marking 22 as property to an identifier of the corresponding bottle 2.

[0143] Bottles 2 having an unacceptable anomaly 14 are preferably discarded, in particular automatically. Depending on the marking procedure, marked or non-marked bottles 2 can be withdrawn / removed directly after anomaly 14 detection or later in the process. Then such bottles 2 can be disposed, cleansed or reused, for example. That is, the one of the bottles 2 under inspection can be marked for rejection if at least one (unacceptable) anomaly 14 is identified.

[0144] The bottle 2 can preferably be thermally sterilized before the optical sensing is performed. The sterilization process has turned out to come along with risks of forming defects or splinters that might even result in a contamination of other bottles 2 without getting a defect. Hence, optically sensing afterwards turned out particularly efficient.

[0145] In one aspect of the present invention, the automatically analyzing is performed using at least one machine learning model 23 which is or has been trained using digital representations 15 of several different sample bottles 2, some of which have and the remaining of which do not have unacceptable anomalies 14, and using pre-known anomaly properties qualifying or classifying each of the sample bottles 2 concerning anomalies 14, preferably as acceptable or unacceptable.

[0146] For the training, the machine learning model 23 is preferably provided with the training data as input data, wherein the machine learning model 23 tries to identify the presence or absence of (unacceptable) anomalies 14, and preferably tries to classify them. Based on the output or the results of the machine learning model 23 and the actual anomalies 14 known for the training data, the machine learning model 23 is then preferably adjusted, in particular the hidden layers, especially the connections or weights between the neurons. By repeating this process, the machine learning model 23 learns to identify and preferably classify anomalies 14 on unknown bottles 2.

[0147] Making use of a machine learning model 23, preferably improved by a self-learning method for improving the machine learning model 23, provides the advantage that anomalies 14 which can be expected in various forms and positions can be more precisely automatically identified, i.e. , found, located, and / or classified.

[0148] The control unit 24 can comprise the machine learning model 23 for finding and / or classifying anomalies 14.

[0149] Preferably, the optically sensing of at least one, preferably an external, surface 12, 13 of the one of the bottles 2 - preferably while content 3 is in the uptake volume 4 - in one aspect of the present invention is limited to a depth of less than 100 micrometer, preferably less than 50 micrometers, in particular 20 micrometer or less, such that at least one of the analyzed digital representations 15 characterizes the surface 12, 13 merely up to said depth. For example, this can be achieved with emission and reflection measurement at a suitable wavelength - depending on the material the bottle 2 is made of- and / or with a suitable angle of emission or detection. Generally speaking, either radiation provided to the bottle 2 or its angle relative to the surface 12, 13 is applied to the bottle 2 such that a penetration depth is limited, or the sensor is configured accordingly or both in combination. Particularly preferably, this is achieved or performed with the dimension sensing device 20 or another sensing device 16 to 19 which is preferably laser based. One or more of the external appearance sensing device 16 and / or the internal appearance sensing device 17 and / or the closure appearance sensing device(s) 18A, 1 SB, 18C and / or the closed bottle sensing device 19 are preferably imaging sensors. Usually, they sense reflected electromagnetic radiation, for example light, from the bottle 2 and convert it to the digital representation 15 being an image or video. However, alternatively or additionally, a transmission or shadow recording can be supported or provided. That is, a source for electromagnetic radiation does not need to be on the same side as the sensing device 16 to 19, but can also be on the opposite side or radiation through the bottle 2 can be reflected or scattered before it is sensed by the sensing device 16 to 19.

[0150] In one aspect of the present invention, the internal surface 13 of the one of the bottles

[0151] 2 is sensed by means of an endoscopic internal appearance sensing device 17 being inserted into the one of the bottles 2. This enables a microscopic image in particular of the bottom 6 and / or a view on areas otherwise covered by the crimp neck 8 or bottle neck.

[0152] The optically sensing (process) can comprise optically sensing the content 3 of the one of the bottles 2 while content 3 is in the uptake volume 4 such that one of the analyzed digital representations 15 characterizes the content 3. Preferably, the content 3 is agitated, in particular whirled, before and / or during sensing the digital representation 15 characterizing the content 3. Sensing the content 3 favorably can be achieved by means of the closed bottle sensing device 19 or a different sensor.

[0153] Alternatively, or additionally, the closed bottle sensing device 19 can be used to generate a digital representation 15 of the closing device 9, in particular the crimp cap 11. In a synergistic manner, the integrity of the closing device 9 or crimp cap 11 on the one hand and analyzing the content 3 for anomalies 14 on the other hand can be performed based on digital representations 15 originating from the same closed bottle sensing device 19, or even the same digital representation 15.

[0154] Optically sensing the content 3 in one aspect of the present invention preferably comprises optically sensing a time series of at least two digital representations 15 characterizing the content 3, and analyzing said digital representations 15 characterizing the content 3 by comparing them with one another. When doing so, the time series of the at least two digital representations 15 characterizing the content

[0155] 3 can be analyzed covering detection of a descent speed of at least one particle. The particle is an anomaly 14 while the speed enables the classification of this anomaly 14. The higher the descent speed the higher is the probability that the particle is not merely an uncritical conglomeration but a foreign substance like a glass splinter. The speed then is used for classification as acceptable or unacceptable anomaly 14.

[0156] The optically sensing (process) alternatively or additionally comprises optically sensing the closing device 9, preferably a septum, that closes off the bottle 2 such that one of the analyzed digital representations 15 characterizes the closing device 9. The closing device 9 can comprise the crimp cap 11 securing the stopper 10 or septum.

[0157] The optically sensing (process) can comprise generating a digital representation 15, preferably based on laser measurement, confocal microscopy, or shadow measurement, of an outer dimension of the one of the bottles 2 such that one of the analyzed digital representations 15 characterizes the outer dimensions.

[0158] Further it may be feasible that the pharmaceutical substance as content 3 is monitored for anomalies 14 prior to filling the one of the bottles 2. For this the monitoring of the content 3, preferably the pharmaceutical substance, may comprise producing one or more digital representations 15 by optically sensing the content 3, and automatically analyzing the one or more digital representations 15 for identifying anomalies 14.

[0159] The method according to the present invention or a packaging process comprising the proposed method of optically inspecting can further comprise a step of filling the one of the bottles 2 or its uptake volume 4 with content 3, preferably pharmaceutical substance, wherein the step of filling the one of the bottles 2 is executed only if no (unacceptable) anomaly 14 of the one of the bottles 2 is identified, in particular when analyzing the digital representation / -s 15 characterizing the internal and / or external surface 12, 13 of the bottle 2 under inspection, and / or wherein the step of filling the one of the bottles 2 is prevented if analyzing the digital representation / -s 15 characterizing the internal and / or external surface 12, 13 of the one of the (empty) bottles 2 results in identifying one or more (unacceptable) anomalies 14. The system 1 preferably comprises a filling appliance 26. The filling appliance 26 preferably comprises the filling station 21 which is configured for filling the content 3 into the bottles 2. Further, the filling appliance 26 preferably comprises at least one appearance sensing device 16, 17, by means of which one or more digital representations 15 can be generated from the empty and open bottle 2. Those digital representations 15 can then be analyzed by means of the analysis module 25.

[0160] In the example shown, the filling appliance 26 comprises at least the external appearance sensing device 16 and the internal appearance sensing device 17. Further, it may be feasible that one, more or all closure appearance sensing devices 18A, 1 SB, 18C form part of the filling appliance 26. Both the external appearance sensing device 16 and the internal appearance sensing device 17 and, preferably, the one, more or all closure appearance sensing devices 18A, 18B, 18C are used to generate digital representations 15 of the external and internal appearance of the same bottle 2, respectively. As already discussed, analysis results can be used separately or in a synergistic combination for identifying anomalies 14, and, preferably, for classifying (qualifying / categorizing) identified anomalies 14 as acceptable or unacceptable.

[0161] If a bottle 2 is found to comprise an unacceptable anomaly 14, this bottle 2 is preferably marked with a marking 22. In the example according to Fig. 1 , the left bottle 2 of the filling station 21 has such marking 22. Preferably, the bottle 2 having such marking 22 is prevented from being filled.

[0162] For some aspects of the present invention, it is sufficient to use only one of the external appearance sensing device 16 or internal appearance sensing device 17 prior to proceeding with the bottle 2 to the filling station 21.

[0163] The filling station 21 can be controlled to provide / fill only bottles 2 without marking 22 with content 3. Of course, there are various options for physically or virtually marking bottles 2 having unacceptable anomalies 14. Marking merely of bottles 2 which do not have unacceptable anomalies 14 is regarded as (inverse) marking of bottles 2 having an unacceptable anomaly 14.

[0164] In one aspect of the present invention, the optically sensing comprises optically sensing a bottle neck, in particular the crimp neck 8, of the bottle 2 while the uptake volume 4 comprises the content 3 after filling, such that at least one of the analyzed digital representations 15 characterizes the appearance of the bottle neck.

[0165] The filling station 21 can comprise the closure appearance sensing device 18A, 1 SB , 18C for checking the bottle 2 in the opening 7 region concerning potential anomalies 14. Preferably, digital representations 15 are generated and analyzed to find / identify and, preferably, to qualify / classify / categorize anomalies 14 as being acceptable or unacceptable. A Bottle 2 comprising one or more unacceptable anomalies 14 is preferably discarded. This can be achieved either by directly withdrawing the bottle 2 or by marking the bottle 2 having at least one anomaly 14 to have an unacceptable anomaly 14 such that it can be withdrawn and discarded afterwards.

[0166] The filling appliance 26 can provide and / or close filled bottles 2 with the closing device 9, preferably with the stopper 10 secured with the crimp cap 11 , e.g., in order to form the vial closed with the septum as used for pharmaceutical purposes. The closure process can be carried out with the bottle closing station C shown schematically in Fig. 1 for clarity and conciseness purposes.

[0167] After closing the filled bottles 2, the closed bottle sensing device 19 can be used to generate one or more digital representations 15 thereof. Alternatively or additionally, the closed and filled bottles 2 can be analyzed by means of the dimension sensing device 20. The closed bottle sensing device 19 and / or the dimension sensing device 20 can form part of a packaging appliance 27. The packaging appliance 27 can be used to add a secondary packaging to the closed and filled bottles 2 for distribution. Alternatively, or additionally, the dimension sensing device 20 can be used to generate a digital representation 15 of the closing device 9, in particular the crimp cap 11 .

[0168] In Fig. 2C to 2F, bottles 2 having different anomalies 14 are depicted. One potential anomaly 14 indicated with an arrow in Fig. 2C is a foreign substance. In Fig. 2D, the bottle 2 has a scratch or crack as anomaly 14. The closing device 9, in particular the crimp cap 11 thereof, can be damaged, as depicted in Fig. 2E, where the damaging can be an anomaly 14 of the bottle 2. Finally, in Fig. 2F, the content 3 comprises a suspended particle indicated with the arrow which can be identified as an anomaly 14. Suspended particles can be qualified as (acceptable or) unacceptable anomalies 14 based on their sensing rate. In particular, pharmaceutical content 3 sometimes comprises acceptable conglomerations which have to be differentiated from particles of foreign substances that are unacceptable anomalies 14. However, usually, the sinking rate of foreign substances is higher than that of uncritical conglomeration. Thus, content 3 of the embodiment shown in Fig. 2F might be the result of agitation steering up the content 3 allowing to determine the sinking rate of the particle being the anomaly 14 in this example, e.g., by taking at least two pictures and determining the sinking rate based on the time period in between. By the outcome, the anomaly 14 can be then qualified, for example by means of the analysis module 25.

[0169] Another aspect of the present invention, which can be realized independently from the other aspects of the present invention and in particular can be applied to different systems in different context as well, relates to a particle detection device 28 for detecting and / or qualifying particles in the content 3 to be filled into the bottles 2. Further, this aspect relates to a system 1 comprising the particle detection device 28, and a method in which particles are detected and / or qualified (to be an anomaly 14 or not) by means of the particle detection device 28.

[0170] The substance, in particular liquid, that is later taken up in the bottle 2 forming its content 3 for simplicity is referred to as content 3 as well even if it is still on its way to the bottle 2, i.e. , in advance of dispensing it into the bottles 2. Thus, the terms “substance that is later taken up in the bottle 2 then forming its content 3” and “content 3” when being outside the bottle are used synonymously and can be replaced with one another.

[0171] The particle detection device 28 is configured and / or used for preferably continuously monitoring a fluid stream of the content 3 before the content 3 is dispensed into the bottles 2.

[0172] Fig. 3 depicts the proposed particle detection device 28 schematically. The particle detection device 28 preferably is inserted in a line 29.

[0173] The line 29 can be configured or used to supply content 3 to the filling station 21 that facilitates or effects dispensing the content 3 into the bottles 2 after the content 3 has passed the particle detection device 28. The particle detection device 28 comprises a window 31 for monitoring the content 3. The particle detection device 28 comprises at least one adapter 30 for adapting the line 29 to the window 31 such that content 3 flowing through the line 29 passes the window 31 in a flow direction f.

[0174] The window 31 is configured such that a particle sensing device 32 and / or a composition sensing device 33 is / are able to detect particles as anomalies 14 in the content 3.

[0175] The window 31 preferably has a geometry that differs from that of the line 29. The line 29 in the example shown is formed by a passage for the substance I content 3 having an essentially circular cross-section. The line 29 can be a hose or tube.

[0176] The window 31 particularly preferably realizes a passage for the content 3 having a flat, in particular essentially rectangular, cross-section. The cross section or a thickness t and a width w of the window 31 in the sense of the present invention is / are perpendicular to a main flow direction f through the window 31.

[0177] The passage for the content 3 realized by the window 31 can have a ratio of width w to thickness t of more than 2:1 or 3:1 , and / or less than 20:1 or 15:1 ; in the example shown about 5:1. This has turned out enabling a good tradeoff between careful guiding the content 3, and reliable particle detection. The resulting thin moving layer of the content 3 through the window 31 improves the detectability of particles even in case of dull contents 3. The window 31 can have a length I (in flow direction f) which is more than 0.5 times the width w and / or less than 5 times the width w.

[0178] In the example shown, a cross-sectional area of the passage of the window 31 is at least essentially the same or similar to that of the line 29 and / or of the adapter 30. Accordingly, the flow speed of the content 3 can essentially be kept constant when flowing through the line 29, the adapter 30, and the window 31 , which preserves the content 3 from being affected due to acceleration, turbulences or the like.

[0179] The (upstream) adapter 30 can taper (decrease the thickness t) in flow direction f and broaden perpendicular to the tapering (increase the width w) in flow direction f, such that the cross-sectional geometry of the passage formed by the line 29 is converted to the cross-sectional geometry of the passage formed by the window 31. A second (downstream) adapter 30 can be provided having the same geometry of its passage, but being passed by the substance I content 3 in the opposite flow direction. That is, the second (downstream) adapter 30 starting from the window 31 can broaden (increase the thickness t) in flow direction f and taper perpendicular to the broadening (decrease the width w) in flow direction f, such that the cross- sectional geometry of the passage formed by the window 31 is converted to the cross-sectional geometry of the passage formed by the line 29.

[0180] The window 31 preferably is at least partially (optically) transparent such that an optical detection is enabled. The particle sensing device 32 and / or the composition sensing device 33 preferably are optical sensing devices like one or more cameras in the example shown. However, the particle sensing device 32 can alternatively or additionally be adapted to find and / or characterize particles by different sensing mechanisms using, e.g., ultrasonic, microwave, terahertz and / or radar technology, while the optical detection has turned out particularly efficient.

[0181] The particle sensing device 32 is configured to detect particles (as or as potential anomalies 14). When the content is flowing through the window 31 , the particle sensing device 32 can optically monitor the content 3 for finding particles. The particle sensing device 32 preferably is or comprises a camera.

[0182] The composition sensing device 33 is configured to analyze the particle for characterizing a substance property. In particular, the composition sensing device 33 is configured to find out whether the particle consists of clumped ingredients of the content 3 or a foreign substance. If the composition sensing device 33 detects a foreign substance particle, it qualifies the particle as anomaly 14. The particle can be qualified as anomaly 14 alternatively or additionally if characterizing the particle results in indicators for the particle being an agglomerate exceeding a density making later dissipation unlikely or the like.

[0183] The composition sensing device 33 preferably records a spectrum or spectral response of the particle and characterizes the particle substance property based thereon. However, the composition sensing device 33 can be configured to taking different measures like contrast detection, shape detection, speed detection, absorbance behavior regarding light, uv, infrared, or millimeter waves, a dielectric property, a radar response, and / or electric conductivity alternatively or additionally. In particular, a spectrum of radiation (light) transmitted, absorbed or reflected, a radiation (light) transmission, absorbance or refection rate, a fluorescent characteristic, and / or a color of the particle can be detected and analyzed by the composition sensing device 33 in order to qualify the particle as anomaly 14 or not. The composition sensing device 33 can use only one or a combination of two or more of the techniques.

[0184] In the example shown, the particle sensing device 32 and the composition sensing device 33 are realized separately. One or more of them can be realized as an optical detector working with backlight emitted by means of an illumination device 34. However, the particle sensing device 32 and the composition sensing device 33 can be realized together, i.e., by the same sensing device as well.

[0185] Results of the particle sensing device 32 and / or the composition sensing device 33 can be evaluated by a processing device 34. In particular, a particle is located by the particle sensing device 32 and a result from the composition sensing device 33 is evaluated at or limited to this location.

[0186] The filling station 21 can comprise multiple lines 29 for filling bottles 2, respectively, and preferably each having a separate particle detection device 28. However, two or more filling openings of the filling station 21 alternatively can share a particle detection device 28. Then, the particle detection device 28 can be followed by a manifold for distributing the substance I content 3 to the bottles 2.

[0187] In the proposed method, the content 4 is conveyed through the window 31 towards the filling appliance 26 by which the content 4 shall be or later is dispensed into the bottles 2. The content 4 actually being in the window 31 is monitored by the particle detection device 28. In particular, the particle sensing device 32 monitors the window 31 to detect particles within the content 4. Optionally, identified particles are examined with the composition sensing device 33.

[0188] The results from the particle sensing device 32 and / or the composition sensing device 33 can be analyzed, e.g., by the processing device 35 in order to qualify the detected particle. For or during analyzation, the features of the particle sensing device 32 and / or the composition sensing device 33 as previously discussed can be used or applied. When a detected particle is qualified as forming an anomaly 14, the process of dispensing the content 4 into the bottles 2 can be stopped or, as preferred, the one of the bottles 2 into which the particle is dispensed is withdrawn, in particular directly or marked to be withdrawn later.

[0189] Different aspects of the present invention can be realized independently from one another and in combination and can provide synergistic advantages in different combinations even though not discussed explicitly for the sake of conciseness.

[0190] Reference Signs:

[0191] System

[0192] Bottle

[0193] Content

[0194] Uptake volume

[0195] Side wall

[0196] Bottom

[0197] Opening

[0198] Crimp neck

[0199] Closing device

[0200] Stopper

[0201] Crimp cap

[0202] External surface

[0203] Internal surface

[0204] Anomaly

[0205] Digital representation

[0206] External appearance sensing device

[0207] Internal appearance sensing device A Opening appearance sensing device B First closure appearance sensing deviceC Second closure appearance sensing device Closed bottle sensing device

[0208] Dimension sensing device

[0209] Filling station

[0210] Marking

[0211] Machine learning model

[0212] Control unit

[0213] Analysis module

[0214] Filling appliance

[0215] Packaging appliance

[0216] Particle detection device

[0217] Line

[0218] Adapter

[0219] Window

[0220] Particle sensing device

[0221] Composition sensing device Illumination device Processing device Transporting means Sterilization equipment Bottle closing station

Claims

Claims:

1. Method of optically inspecting bottles (2), in particular vials, each having an uptake volume (4) for storing content (3), preferably medical products, bounded by an internal surface (13) of the bottle (2), the uptake volume (4) being accessible via an opening (7) of the bottle (2), the method comprising for each of several bottles (2) under inspection: producing digital representations (15) of one of the bottles (2) by optically sensing the one of the bottles (2), and automatically analyzing the digital representations (15) for identifying anomalies (14), wherein the optically sensing comprises optically sensing an external appearance of the one of the bottles (2), in particular while the uptake volume (4) is empty, such that at least one of the analyzed digital representations (15) characterizes the external appearance, and wherein the optically sensing comprises optically sensing an internal appearance at least of a bottom (6) of the one of the bottles (2) through the opening (7) while the uptake volume (4) is empty such that at least one of the analyzed digital representations (15) characterizes the internal appearance of the bottom (6), wherein two or more different digital representations (15) of the same bottle (2) or a part thereof are analyzed together in order to identify anomalies (14).

2. Method according to claim 1 , characterized in that the method comprises identifying an anomaly (14) by finding discrepancies between different digital representations (15) of the same region, in particular a structure that is represented differently in the digital representations (15).

3. Method according to claim 1 , characterized in that the method comprises marking the one of the bottles (2) for rejection if at least one anomaly (14) is identified.

4. Method according to claim 1 or 2, characterized in that the method comprises thermally sterilizing the one of the bottles (2) before the optical sensing is performed.

5. Method according to one of the preceding claims, characterized in that the automatically analyzing is performed using at least one machine learning model (23) which is trained using digital representations (15) of several different6. sample bottles (2), some of which having and the remaining of which not having anomalies (14), and using pre-known anomaly properties qualifying each of the sample bottles (2) concerning anomalies (14).

7. Method according to one of the preceding claims, characterized in that the optically sensing comprises optically sensing at least one, preferably an external, surface of the one of the bottles (2) - preferably while content (3) is in the uptake volume (4) - up to a depth of less than 100 micrometer, preferably less than 50 micrometers, in particular 20 micrometer or less, such that at least one of the analyzed digital representations (15) characterizes the surface merely up to said depth.

8. Method according to one of the preceding claims, characterized in that the optically sensing the internal appearance of the one of the bottles (2) comprises using an endoscopic sensing device being inserted into the one of the bottles (2).

9. Method according to one of the preceding claims, characterized in that, while content (3) is in the uptake volume (4), the optically sensing comprises optically sensing the content (3) of the one of the bottles (2) such that one of the analyzed digital representations (15) characterizes the content (3), preferably wherein the content (3) is agitated, in particular whirled, before and / or during sensing the digital representation (15) characterizing the content (3).

10. Method according to claim 7, characterized in that optically sensing a content (3) comprises optically sensing a time series of at least two digital representations (15) characterizing the content (3), and analyzing the digital representations (15) characterizing the content (3) by comparing them with one another, preferably wherein analyzing the time series of the at least two digital representations (15) characterizing the content (3) covers detection of a descent speed of at least one particle.

11. Method according to one of the preceding claims, characterized in that the optically sensing comprises optically sensing a closing device (9), preferably a septum, that closes off the bottle (2) such that one of the analyzed digital representations (15) characterizes the septum.

12. Method according to one of the preceding claims, characterized in that the optically sensing comprises optically sensing an outer dimension of the one of the bottles (2) such that one of the analyzed digital representations (15) characterizes the outer dimensions, wherein the optically sensing is based on laser measurement, confocal microscopy, or shadow measurement.

13. Method according to one of the preceding claims, further comprising a step of filling the one of the bottles (2) with content (3), preferably a pharmaceutical substance, wherein the step of filling the one of the bottles (2) is executed only if no anomaly (14) of the one of the bottles (2) is identified when analyzing the digital representation's characterizing the internal and / or external surface (12, 13) of the bottle (2) under inspection, and wherein the step of filling the one of the bottles (2) is prevented if analyzing the digital representation / -s (15) characterizing the internal and / or external surface (12, 13) of the one of the bottles (2) results in identifying one or more anomalies (14).

14. Method according to claim 11 , characterized in that the optically sensing comprises optically sensing a bottle neck, in particular crimp neck (8), of the bottle(2) while the uptake volume (4) comprises the content (3) after filling, such that at least one of the analyzed digital representations (15) characterizes the appearance of the bottle neck.

15. Method according to one of the preceding claims, characterized in that content (3) is monitored by means of a particle detection device (28) for detecting and / or qualifying particles in the content (3) before entering the bottles (2), preferably: wherein the particle detection device (28) comprises a line (29) for supplying content(3) to a filling station (21 ) and a window (31 ) for monitoring the content (3), wherein the particle detection device (28) comprises at least one adapter (30) for adapting the line (29) to the window (31) such that content (3) flowing through the line (29) passes the window (31) in a flow direction (f), preferably wherein the window (31 ) has a geometry that differs from a geometry of the line (29), in particular wherein the window (31 ) realizes a passage for the content (3) having a flat cross-section, and / or wherein the particle detection device (28) detects particles in the content (3) using a particle sensing device (32) and / or qualifies particles to be anomalies (14) based ona composition of the detected particles detected using a composition sensing device (33).

16. System (1) for optical inspection of bottles (2) for storing medical products, in particular vials, comprising: one or more optical sensing devices for optically sensing one of the bottles (2) under inspection for producing digital representations (15) by optically sensing the one of the bottles (2), an analysis module (25) for automatically analyzing the one or more digital representations (15) for identifying anomalies (14), wherein the system (1) is configured to execute the method according to one of claims 1 to 12.

17. Computer program comprising instructions to cause the system (1 ) of claim 13 to execute the steps of the method according to one of claims 1 to 12.

18. Computer-readable medium having stored thereon the computer program of claim 14.

Images