Method for detecting an anomaly in a strip of semi-finished rubbery product

Abstract

A device is proposed for detecting at least one anomaly in a surface area (7) of a strip (2) of a semi-finished rubber product. The device comprises: - acquisition means (8) for capturing an image of the surface area (7), - a memory (10) containing computer program instructions (11), and - a processing unit (15) configured to execute the computer program instructions (11) stored in the memory (10). The computer program instructions (11) are configured, when executed, to: - obtain an image of the surface area of ​​the strip, - implement an assignment function to assign an anomaly score to the image, - compare the anomaly score of the image to at least one predetermined positive detection threshold using comparison means (12), and - detect the anomaly in the surface area (7) based on the result of the comparison. Figure for the abstract: Fig. 1

Description

Title of the invention: Method for detecting an anomaly in a strip of semi-finished rubbery product. Technical field

[0001] The present invention relates to the field of visual anomaly detection in a semi-finished rubbery product used in the manufacture of tires.

[0002] More specifically, the invention relates to a method for detecting an anomaly in a surface area of ​​a strip of semi-finished rubbery product, a device for detecting an anomaly in such a surface area and a manufacturing line comprising such a device. Context

[0003] Generally, a tire is obtained by curing (vulcanizing) an assembly of strips of at least one semi-finished rubbery product and layers of yarn.

[0004] The rubbery semi-finished product strips result from a mixing process implemented by a mixing line.

[0005] During the mixing process, raw materials comprising an elastomer, a reinforcing filler and additives including sulfur are mixed in predefined proportions and order, and are heated to a predefined temperature to obtain a homogeneous rubbery mixture.

[0006] An extruder fed by the rubber mixture shapes a continuous strip of the semi-finished rubber product.

[0007] The continuous strip of the semi-finished rubbery product is inspected to detect defects such as the presence of foreign bodies in the strip.

[0008] Then the strip is divided into plates for storage before being used for tire manufacturing.

[0009] Document CN219285437 discloses a device for detecting a metallic foreign body in a rubber band.

[0010] Document JPH01228805 discloses a device for detecting a foreign body in rubber plates, the foreign body being in particular metallic.

[0011] However, during the mixing of the raw materials, foreign bodies of plastic, wood or paper may be mixed with the raw materials.

[0012] Foreign bodies are generally pieces of raw material containers.

[0013] Since known prior art detection devices only detect metallic foreign bodies, a rubber plate containing at least one non-metallic foreign body can be declared compliant.

[0014] Furthermore, the detection devices known in the prior art do not allow the detection of an agglomerate of additives in the strip resulting from a heterogeneous mixture of raw materials so that a rubber plate made from a heterogeneous mixture of raw materials can be declared compliant.

[0015] Tires made from such plates are not compliant and are rejected, inducing significant non-compliance costs.

[0016] Generally, a strip made from a heterogeneous mixture has traces of yellow color on its surface, the yellow traces indicating an excessive presence of sulfur.

[0017] To ensure that the declared compliant rubber plates can be used to manufacture compliant tires, it is necessary to be able to detect the presence of non-metallic foreign bodies in the rubber plates and to identify the plates made from a heterogeneous mixture.

[0018] An operator can perform a visual inspection of the surface of the strip to detect the presence of a metallic or non-metallic foreign body and / or traces of yellow color.

[0019] However, visual inspection is a tedious task. Furthermore, the reliability of visual inspection by the operator depends on the operator's attention and experience.

[0020] It is proposed to overcome all or part of the known drawbacks of the prior art. Summary of the invention

[0021] The present invention relates to a device for detecting at least one anomaly in a surface area of ​​a strip of semi-finished rubbery product.

[0022] The device comprises:

[0023] - acquisition means for capturing an image of the surface area of ​​the strip of semi-finished product,

[0024] - a memory containing computer program instructions, and

[0025] - a processing unit configured to execute program instructions computer program instructions are stored in memory and configured to execute the following when they are run:

[0026] - obtain an image of the surface area of ​​the strip,

[0027] - implement an assignment function to assign an anomaly score to the image, the value of the anomaly score being representative of the degree of dissimilarity from the content of the image to the content of at least one reference image that does not contain any anomalies,

[0028] - compare the image anomaly score to at least one positive detection threshold predetermined by means of comparison, and

[0029] - detect the anomaly in the surface area from the result of the comparison.

[0030] The detection device makes it possible to systematically and in real time analyze images of surface areas of the rubbery semi-finished product strip to detect the presence of at least one metallic, non-metallic foreign body and traces of an additive having a color different from the color palette of the strip representative of a heterogeneous mixture of the raw materials of the semi-finished product strip.

[0031] The reliability of detecting an anomaly in the semi-finished product strip is constant and improved compared to known prior art detection devices.

[0032] Preferably, the anomaly is detected when the anomaly score is greater than the detection threshold.

[0033] Advantageously, the anomaly comprises an inclusion of a non-metallic foreign body in the surface area and / or at least one trace of color different from a predetermined set of colors of the semi-finished product strip made from a homogeneous mixture of raw materials.

[0034] Preferably, the assignment function includes an artificial neural network.

[0035] Advantageously, the artificial neural network is a convolutional artificial neural network.

[0036] Preferably, the device further comprises display means configured to represent each area of ​​the image dissimilar to the content of the reference image by a first set of predetermined colors and each area of ​​the image similar to the content of the reference image by a second set of predetermined colors different from the colors of the first set of colors.

[0037] The invention further relates to a control line for a strip of semi-finished rubbery product.

[0038] The control line comprises:

[0039] - a conveyor configured to convey the semi-finished product belt,

[0040] - acquisition means configured to capture an image of the surface area from the semi-finished product strip,

[0041] - a detection device as defined above,

[0042] - a gantry arranged above the conveyor, the acquisition means being fixed on the gantry and oriented to capture the surface area, and

[0043] - means for measuring the unwinding distance of the semi-finished product strip finished.

[0044] The invention further relates to a method for detecting at least one anomaly in a surface area of ​​a strip of semi-finished rubbery product implemented by computer.

[0045] The process comprises:

[0046] - obtaining an image of the surface area of ​​the strip,

[0047] - an implementation of an assignment function for assigning an anomaly score In the image, the anomaly score value is representative of the degree of dissimilarity between the image content and the content of at least one reference image that does not contain anomaly.

[0048] - a comparison of the image anomaly score to at least one detection threshold positive predetermined by means of comparison, and

[0049] - an anomaly detection in the surface area from the result of the comparison.

[0050] The invention relates to a computer program comprising instructions which, when the program is executed by a computer, lead the latter to implement the steps of the detection process as defined above. Brief description of the drawings

[0051] The present invention will be better understood and other objects, advantages and features will become apparent from the detailed description that follows, including embodiments given by way of illustration only and made with reference to the accompanying drawings, presented as non-limiting examples, which may serve to complete the understanding of the invention and the explanation of its implementation and, where appropriate, contribute to its definition, on which:

[0052] [Fig-1] Fig. 1 schematically illustrates a control line of a strip of semi-finished rubbery product.

[0053] [Fig.2] [Fig.2] schematically illustrates an example of a detection method of at least one anomaly in a surface area of ​​the strip of semi-finished rubbery product.

[0054] [Fig.3] Fig.3 schematically illustrates an example of the surface area of ​​the band.

[0055] [Fig.4] Figure 4 schematically illustrates an example of an anomaly map of the surface area of ​​the rubbery semi-finished product strip of [Fig.3].

[0056] [Fig. 5] Figure 5 schematically illustrates an example of a training method of an artificial neural network. Detailed description

[0057] Reference is made to [Fig.1] which schematically illustrates a control line 1 of a strip of semi-finished rubbery product 2.

[0058] The semi-finished rubbery product strip 2 is obtained by the continuous extrusion of a mixture of elastomer, a reinforcing filler and additives including a crosslinking system.

[0059] The elastomer is chosen from the group consisting of diene elastomers and mixtures thereof. The term "diene" elastomer (or "rubber," the two terms being considered synonymous) is understood here to mean, in a known manner, one (or more) elastomers derived at least in part (i.e., a homopolymer or a copolymer) from diene monomers (monomers bearing two carbon-carbon double bonds, conjugated or not). Any type of reinforcing filler known for its ability to strengthen a rubber composition suitable for tire manufacturing may be used; for example, an organic filler such as carbon black, a reinforcing inorganic filler such as silica or alumina, or a mixture of these two types of filler. Preferably, the reinforcing load ratio is in the range of 5 to 200 pc, preferably 20 to 160 pc.The reinforcing filler is preferably chosen from the group consisting of silicas, carbon blacks, and mixtures thereof. More preferably, the reinforcing filler is predominantly carbon black, preferably at a concentration in the range of 30 to 90 parts per cent. Also preferably, the reinforcing filler is predominantly silica, preferably at a concentration in the range of 30 to 90 parts per cent.

[0060] Any type of crosslinking system known to those skilled in the art may be used for rubber compositions. Preferably, the crosslinking system is a vulcanizing system, i.e., based on sulfur (or a sulfur-donating agent) and a primary vulcanizing accelerator. Various known secondary accelerators or vulcanizing activators, such as zinc oxide, stearic acid or equivalent compounds, and guanidine derivatives (in particular diphenylguanidine), may be added to this basic vulcanizing system and incorporated during the first non-productive phase and / or during the productive phase as described later. Sulfur is used at a preferential rate of between 0.5 and 10 parts per million (ppm), more preferably between 0.5 and 5 ppm, and in particular between 0.5 and 3 ppm.

[0061] The term "pce" means, for the purposes of this patent application, part by weight per hundred parts of elastomers, as defined in the preparation of the composition before baking.

[0062] The mixing is carried out by a calibrated mixing process to introduce the components of the mixture in defined proportions and in a defined order according to a recipe, and to mix the products while respecting at least one manufacturing parameter This process may include, for example, a duration, a set temperature, a quantity of energy supplied to the mixture, or a combination of several parameters. The mixing process homogenizes and kneads the mixture according to the recipe.

[0063] The strip of semi-finished rubbery product 2 resulting from the extrusion is suitable for shaping, for example to form a tire and for vulcanization.

[0064] The mixing process and the extrusion process are implemented by a mixing line (not shown) comprising at least one mixer and at least one extruder.

[0065] The control line 1 includes a conveyor 3 comprising a conveyor belt 3a, a gantry 4 arranged above the conveyor belt 3a.

[0066] The semi-finished product strip 2 is placed on the conveyor belt 3a to be conveyed to a cutting station capable of cutting the semi-finished product strip 3a into portions of semi-finished product strip intended to be stacked.

[0067] The conveyor 3 can for example convey the semi-finished product strip 2 from the exit of the extruder to the cutting station.

[0068] The control line 1 further comprises a gantry 4 arranged above the conveyor belt 3a, lighting means 5, means for measuring the unwinding distance of the semi-finished product strip and a device for detecting at least one anomaly in a surface area 7 of the semi-finished product strip 2.

[0069] Of course, the surface area 7 of the semi-finished product strip 2 may contain a plurality of anomalies and the detection device may detect said plurality of anomalies of said surface area 7.

[0070] The width of zone 7 is at least equal to the width 12 of the semi-finished product strip 2 and the length L2 of the surface zone 7 is equal to a predetermined capture length.

[0071] An anomaly includes, for example, an inclusion of a foreign body in zone 7.

[0072] The anomaly may include a trace of color different from a predetermined set of colors of the semi-finished product strip made from a homogeneous mixture of raw materials.

[0073] The foreign body may include a metallic or non-metallic body.

[0074] The non-metallic foreign body includes, for example, a wooden foreign body, plastic or paper.

[0075] The foreign body comes for example from a container of the elastomer, the filler or an additive.

[0076] The predetermined colors of the semi-finished product strip are those of the semi-finished product strip when the raw material mixture is homogeneous, particularly when the strip of semi-finished product does not contain any traces of yellow color.

[0077] The lighting means 5 include, for example, light-emitting diodes.

[0078] The measuring means 6 include, for example, an encoder wheel.

[0079] The detection device includes acquisition means 8, and a processing device 9 connected to the acquisition means 8 and to the measurement means 6.

[0080] The acquisition means 8 include, for example, a color camera.

[0081] The camera is, for example, of the 4K type.

[0082] The acquisition means 8 are oriented so as to capture an image of the surface area 7 of the band 2 illuminated by the lighting means 5.

[0083] The image captured by the acquisition means 8 comprises a plurality of pixels forming said image.

[0084] Alternatively, if the ambient lighting of the surface area 7 of the semi-finished product strip 2 is sufficient to capture an image of the surface area 7 by the acquisition means 8, the lighting means 5 are not activated or the control line 1 may not include lighting means 5.

[0085] The gantry 4 can be fixed or mobile so that it moves in a direction parallel to the conveyor belt 3a.

[0086] The processing device 9 includes a memory 10 storing a computer program 11, comparison means 12, display means 13 and a processing unit 14.

[0087] The processing unit 14 is configured to implement the computer program 11 stored in memory 10, the comparison means 12 and the display means 13.

[0088] The display means 14 include, for example, a viewing screen.

[0089] Fig. 2 illustrates an example of a method for detecting at least one anomaly in the surface area 7 of the strip 2 of semi-finished rubbery product.

[0090] The detection device carries out the detection process.

[0091] The computer program 11 includes instructions which, when the program is executed by the processing unit 14, lead the latter to implement the steps of said detection process described below.

[0092] During a step 20, the acquisition means 8 capture an image of the surface area 7.

[0093] The image comprises a plurality of pixels forming the image.

[0094] During a step 21, an assignment function assigns an anomaly score to the image captured by the acquisition means 8.

[0095] The anomaly score is representative of the degree of dissimilarity of the content of the captured image to the content of reference images not containing anomaly and allows us to assess the probability of the presence of at least one anomaly in a region of the captured image.

[0096] The assignment function extracts target features from the image and compares the target features of the image to reference features stored in memory 10.

[0097] The reference features are previously extracted (learning step) from at least one reference image of a surface area of ​​the semi-finished product strip 2 showing no anomalies by the assignment function.

[0098] The surface area of ​​the semi-finished product strip 2, which is free of anomalies, does not contain any metallic or non-metallic foreign bodies, and the strip does not show any trace of yellow color representative of a heterogeneous mixture of the raw materials constituting the surface area of ​​the strip.

[0099] The assignment function assigns to each pixel of the image captured in step 20 a positive anomaly value based on the result of comparing the target image features to the reference features.

[0100] The image anomaly score is equal to the maximum anomaly value of all pixels in the image captured in step 20.

[0101] During a step 22, the comparison means 12 compare the image anomaly score to a predetermined detection threshold.

[0102] The predetermined detection threshold is a positive value and is determined, for example, by analyzing the distribution of anomaly scores in a set of reference images. The threshold is chosen, for example, such that a predetermined percentage, for example equal to 95% or 99%, of the anomaly scores in the reference images are below this threshold.

[0103] If the anomaly score is below the detection threshold, the image content is considered to be anomaly-free. Surface area 7 of band 2 is anomaly-free and therefore conforms.

[0104] Conversely, if the image anomaly score is greater than the detection threshold, the image content is considered to contain at least one anomaly. Surface area 7 of band 2 is not compliant.

[0105] The display means 13 can display on the viewing screen an anomaly map in which each pixel in each region of the captured image dissimilar to the content of the reference images is represented by a first set of predetermined colors and each pixel in each region of the image similar to the content of the reference images is represented by a second set of predetermined colors different from the colors of the first set.

[0106] The color assigned to each pixel is mapped to one of the colors of the first or second set of predetermined colors according to the anomaly value of said pixel. This correspondence can, for example, be performed via linear interpolation in the color space formed by the colors of the first and second predetermined color sets, with the scores being converted into RGB values.

[0107] For example, the blue color of the second set of predetermined colors represents pixels without anomalies, the orange color of the first set represents pixels likely to have at least one anomaly, and the red color of the first set represents pixels with at least one anomaly.

[0108] The anomaly map allows an operator to visualize the location of the anomaly on the surface area 7.

[0109] The pixel anomaly values ​​can be smoothed and normalized, for example, by applying Gaussian filtering prior to obtaining the anomaly map to soften the edges of the anomaly map and make the interpretation of the anomaly map easier.

[0110] Furthermore, the threshold can be applied to the anomaly map to segment the abnormal regions in the image. Pixels whose anomaly value exceeds the threshold are colored according to one of the colors from the first set of colors, allowing the operator to precisely visualize the suspect regions of the anomaly map and the corresponding surface area 7.

[0111] The assignment function includes, for example, a convolutional artificial neural network configured to assign an anomaly score to the image captured by the acquisition means 8 and a learnable patch descriptor arranged as described in the document entitled "CFA: Coupled-Hypersphere-Based Feature Adaptation for Target-Oriented Anomaly localization", SUNGWOOK LEE, SEUNGHYUN LEE, AND BYUNG CHEOL SONG, published on July 25, 2022, IEEE.

[0112] The trained artificial neural network makes it possible to extract target features from the captured image and the learnable patch descriptor makes it possible to transform the extracted target features into a form adapted specifically to the training dataset used during the training of the artificial neural network in order to reduce the potential biases present in the target features extracted by the artificial neural network.

[0113] Now, an example of a method for controlling band 2 implementing control line 1 is described.

[0114] The conveyor belt 3a is continuously driven by drive means (not shown) so that the belt 2 unwinds and passes under the fixed gantry 4.

[0115] When the band 2 has traveled the capture length, the detection process is implemented for the first time.

[0116] Steps 20 to 22 are implemented by the detection device.

[0117] If the anomaly score exceeds the detection threshold, the display means 13 display the anomaly card. The processing device 9 activates, for example, a human-machine interface to notify an operator of the detection of an anomaly. The human-machine interface includes, for example, a display unit.

[0118] Following the viewing of the anomaly card, the operator can stop conveyor 3 to extract the part of belt 2 containing the anomaly.

[0119] Alternatively, the portion of belt 2 containing the anomaly is identified, and when this portion of belt 2 is cut into at least one section of semi-finished product, said section of belt is discarded. Conveyor 3 is not stopped, so the mixing line and the control line are not stopped.

[0120] The distance measuring means measure the distance traveled by the conveyor belt 5a which is equal to the unwinding distance of the semi-finished product belt 2.

[0121] When strip 2 has traveled the predetermined capture length from the image capture in step 20 during the first implementation of the detection method, the detection method is implemented a second time. Steps 20 to 22 are repeated to detect an anomaly in another surface area of ​​strip 2.

[0122] The detection process is repeated as soon as the band 2 has traveled the predetermined capture distance from the image capture by the acquisition means 8 of the implementation of the previous detection process.

[0123] When the gantry 4 is mobile, it can be moved on the control line and immobilized during the implementation of the detection process.

[0124] Fig. 3 illustrates an example of surface area 7 of band 2.

[0125] Foreign bodies 30, 31 are present in surface area 2.

[0126] Foreign bodies 30, 31 are for example made of wood.

[0127] Surface area 7 of the band has two anomalies.

[0128] Figure 4 illustrates an example of the anomaly map produced by the process of detection from the image of the surface area 7 of band 2 illustrated in [Fig.3].

[0129] The anomaly map has two portions 40, 41 represented by the first set of colours signaling the presence of foreign bodies 30, 31.

[0130] A first portion 40 comprises a first zone 40a colored in a first color from the first set of colors, for example in yellow, a second zone 40b colored in a second color from the first set of colors, for example in light orange, a third zone 40c colored in a third color from the first set of colors, for example in dark orange and a fourth zone 40d colored in a fourth color from the first set of colors, for example in red.

[0131] The first color represents a first probability of the presence of an anomaly, the second color represents a second probability of the presence of an anomaly greater than the first probability of the presence of an anomaly, the third color represents a third probability of the presence of an anomaly greater than the first and second probabilities of the presence of an anomaly, and the fourth color represents the presence of an anomaly.

[0132] Similarly, the second portion 41 comprises a first zone 41a coloured according to the first colour of the first set of colours, a second zone 41b coloured according to the second colour of the first set of colours, a third zone 41c coloured according to the third colour of the first set of colours and a fourth zone 41d coloured according to the fourth colour of the first set of colours.

[0133] Portions 42 of the anomaly map similar to the content of at least one reference image are represented by at least one color from the second color set, for example in blue.

[0134] Fig. 5 illustrates an example of a method for training the assignment function.

[0135] During a step 50, images of the semi-finished product strips are captured for example by the acquisition means 8 to form a training dataset.

[0136] The strips of semi-finished product selected for obtaining the training dataset do not exhibit any anomalies. No metallic or non-metallic foreign matter is present in the selected strips, and the selected strips show no trace of yellow color indicative of a heterogeneous mixture of the raw materials constituting said strips.

[0137] Each image in the training dataset is a reference image.

[0138] During step 51, the artificial neural network receives as input each image from the training dataset and extracts the target features.

[0139] The patch descriptor transforms the extracted target features into a form specifically adapted to the dataset and stores in memory 10 the reference features which include the extracted and transformed target features.

[0140] In addition, the assignment function assigns an anomaly score to said image.

[0141] The patch descriptor is trained to bring the target features closer to the reference features previously stored in memory 10 to create hyperspheres around the reference features in order to facilitate the distinction between the target features from an image with an anomaly and the target features from an image without an anomaly.

[0142] During a step 52, the comparison means 11 compare the anomaly score to the predetermined threshold to determine whether the anomaly score is above or below said threshold. Since the images in the training dataset do not contain any anomalies, during a step 33, weights of the artificial neural network are adjusted so that the anomaly score of each image is below the detection threshold.

[0143] The detection device makes it possible to analyze in real time images of surface areas of the rubbery semi-finished product strip to detect the presence of at least one metallic, non-metallic foreign body and traces of an additive having a color different from the color palette of the strip representative of a heterogeneous mixture of the raw materials of the strip unlike the detection devices known in the prior art which make it possible to detect metallic foreign bodies.

[0144] The reliability of detecting an anomaly in band 2 of the detection device is improved compared to known prior art detection devices.

[0145] Furthermore, the operator, alerted by the human-machine interface, intervenes only on a limited number of portions of tape 2 containing at least one anomaly, so that the operator's attention is focused on a portion of tape 2 containing at least one anomaly. The detection device helps to reduce the operator's mental workload.

[0146] Of course, the anomaly detection method can be implemented to detect an anomaly in a surface area of ​​a strip of semi-finished rubbery product resulting from a process other than a mixing process and an extrusion process, for example a calendering process.

Claims

Demands

1. Device for detecting at least one anomaly in a surface area (7) of a strip (2) of a semi-finished rubber product, the device comprising: - acquisition means (8) for capturing an image of the surface area (7) of the strip (2) of the semi-finished product, - a memory (10) containing computer program instructions (11), and - a processing unit (15) configured to execute the computer program instructions (11) stored in the memory (10), the computer program instructions (11) being configured to, when executed: - obtain an image of the surface area (7) of the strip (2), - implement an assignment function to assign an anomaly score to the image, the value of the anomaly score being representative of the degree of dissimilarity of the image content to the content of at least one reference image not containing an anomaly,- compare the image anomaly score to at least one predetermined positive detection threshold using comparison means (12), and - detect the anomaly in the surface area (7) from the comparison result.

2. Device according to claim 1, wherein the anomaly is detected when the anomaly score is greater than the detection threshold.

3. Device according to claim 1 or 2, wherein the anomaly comprises an inclusion of a non-metallic foreign body in the surface area and / or at least one trace of color different from a predetermined set of colors of the strip (2) of semi-finished product made from a homogeneous mixture of raw materials.

4. Device according to any one of claims 1 to 3, wherein the assignment function comprises an artificial neural network.

5. Device according to claim 4, wherein the artificial neural network is a convolutional artificial neural network.

6. A device according to any one of claims 1 to 9, further comprising display means (13) configured to represent each area of ​​the image dissimilar to the content of the reference image by a first set of predetermined colors and each area of ​​the image similar to the content of the reference image by a second set of predetermined colors different from the colors of the first set of colors.

7. Control line (1) of a rubbery semi-finished product strip (2), comprising: - a conveyor (3) configured to convey the semi-finished product strip (2), - acquisition means (8) configured to capture an image of the surface area (7) of the semi-finished product strip (2), - a detection device according to any one of claims 1 to 6, - a gantry (4) disposed above the conveyor (3), the acquisition means (8) being fixed on the gantry and oriented to capture the surface area, and - means for measuring the unwinding distance of the semi-finished product strip.

8. A computer-implemented method for detecting at least one anomaly in a surface area (7) of a strip (2) of semi-finished rubber product, the method comprising: - obtaining an image of the surface area (7) of the strip (5), - implementing an assignment function to assign an anomaly score to the image, the value of the anomaly score being representative of the degree of dissimilarity of the content of the image to the content of at least one reference image not containing an anomaly, - comparing the anomaly score of the image to at least one predetermined positive detection threshold by comparison means (12), and - detecting the anomaly in the surface area (7) from the result of the comparison.

9. Computer program (11) comprising instructions which, when the program (11) is executed by a computer, cause the computer to carry out the steps of the detection method according to claim 8.

Images