Foreign matter detection method, foreign matter detection system and program

Abstract

To accurately detect a foreign matter in liquid by imaging a container storing liquid, setting a plurality of regions in a photographed image of the container, and efficiently removing noise by changing a noise removal capability of a filter to be applied according to a portion of the container and a time zone in which the container is imaged.SOLUTION: A foreign matter detection method comprises: an imaging process of imaging a container made of a translucent material storing liquid; an image processing process of performing filter processing on the photographed image of the container; and a detection process of detecting a foreign matter contained in the container on the basis of the image subjected to the filter processing. The image processing process sets a plurality of inspection regions on the photographed image of the container, sets the plurality of time zones on the basis of the timing at which the container is imaged, and sets a filter having a noise removal capability in a plurality of inspection regions for each of the plurality of time zones.SELECTED DRAWING: Figure 13

Description

[Technical field]

[0001] The present invention relates to a method, a system, and a program for executing the method for detecting foreign matter in a liquid contained in a container. [Background technology]

[0002] A container filled with liquid may contain foreign matter, and therefore it is necessary to inspect the liquid for foreign matter. In addition, when inspecting, a reference value must be set so that air bubbles in the liquid are not erroneously detected as foreign matter, which may result in minute foreign matter not being detected. In order to avoid such a situation, an inspection device has been disclosed that divides a container into a number of areas, sets a reference value for each area, and when inspecting each area for foreign matter, compares the reference value for that area to improve the detection of foreign matter (see, for example, Patent Document 1).

[0003] Furthermore, conventionally, a device for inspecting whether a foreign object is present in a liquid contained in a container has been developed that rotates the container to move the foreign object in the container and captures an image of the moving object to detect it as a foreign object. For example, an inspection system has been disclosed that includes an inversion device that tilts or inverts the container and returns it to its original position, and an imaging device that captures an image of the container immediately after it has been tilted or inverted by the inversion device and returned to its original position (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 3-96841 [Patent Document 2] International Publication No. 2005 / 031328 Summary of the Invention [Problem to be solved by the invention]

[0005] However, with an inspection device such as that disclosed in Patent Document 1, it is difficult to accurately distinguish between foreign bodies and air bubbles and detect only foreign bodies, when the air bubbles change over time across each set area. Also, when using an inspection system that detects foreign bodies by inverting a container and returning it to its original position, such as the inspection system disclosed in Patent Document 2, air bubbles are generated by inverting the container and returning it to its original position, and therefore foreign bodies cannot be detected accurately unless the air bubbles are efficiently removed.

[0006] Therefore, the foreign object detection method of the present invention aims to provide a foreign object detection method that reduces the rate of false detection of foreign objects by photographing a container containing liquid, setting multiple regions in the photographed image of the container, and efficiently removing noise by changing the noise removal capability of the applied filter depending on the part of the container and the time of day the container was photographed.

[0007] Furthermore, the foreign object detection method of the present invention aims to provide a foreign object detection method that enables foreign objects to be detected with high accuracy by photographing a container while being turned upside down, thereby moving the foreign object within the liquid, while efficiently removing noise such as air bubbles caused by turning the container upside down by changing the noise removal capability of the applied filter depending on the part of the container and the time of day the container is photographed, thereby reducing the rate of false detection of foreign objects. [Means for solving the problem]

[0008] In order to achieve the above object, a foreign object detection method according to an embodiment of the present invention includes: An imaging process for imaging a container made of a light-transmitting material containing a liquid; an image processing process for filtering the captured image of the container; a detection process for detecting foreign objects contained within the container based on the filtered image; The image processing process is characterized in that it sets a plurality of inspection areas in the image of the container, sets a plurality of time periods based on the timing at which the image of the container was taken, and sets a filter having noise removal capability in the plurality of inspection areas for each of the plurality of time periods.

[0009] Moreover, the foreign object detection system according to the embodiment of the present invention includes: A means for turning upside down a container made of a light-transmitting material containing a liquid; A means for photographing the container; A means for processing the captured image of the container, The means for processing the image includes: A region setting means for setting a plurality of regions in an image of the container; an image processing means for filtering the image by setting a filter having a noise removal capability for the plurality of regions in accordance with the posture of the photographed container; and a foreign matter detection means for detecting foreign matter contained in the liquid based on the filtered image.

[0010] In addition, the computer program according to the embodiment of the present invention includes an imaging step of imaging a container made of a light-transmitting material containing a liquid; an image processing step of filtering the captured image of the container, the image processing step including: setting a plurality of inspection areas in the image of the container; setting a plurality of time periods based on the timing at which the image of the container was captured; and setting a filter having noise removal capability in the plurality of inspection areas for each of the plurality of time periods; and a detection step of detecting foreign matter contained in the container based on the filtered image. Effect of the Invention

[0011] As described above, according to the invention disclosed herein, a container containing liquid is photographed, multiple regions are set in the photographed image of the container, and the noise removal capability of the applied filter is changed depending on the part of the container and the time of day the container is photographed, thereby efficiently removing noise, thereby making it possible to detect foreign objects with a reduced rate of false positives. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a front view of a foreign object detection system according to an embodiment. [Diagram 2] FIG. 2 is a side view of the foreign object detection system shown in FIG. [Diagram 3] FIG. 3 is a cross-sectional view of a bottle (container) to be placed on the foreign object detection system shown in FIG. 1. [Figure 4] FIG. 4 is a diagram showing a state in which the inclination of the bottle changes when a foreign object is detected according to the embodiment. [Diagram 5] FIG. 5 is a diagram showing a state in which the inclination of the bottle changes during the photographing process of the foreign object detection method according to the embodiment. [Figure 6A] FIG. 6A is a diagram illustrating the movement of air and foreign matter during the imaging process of the foreign matter detection method according to the embodiment. [Figure 6B] FIG. 6B is a diagram illustrating the movement of air and foreign objects during the imaging process of the foreign object detection method according to the embodiment. [Figure 6C] FIG. 6C is a diagram illustrating the movement of air and foreign matter during the imaging process of the foreign matter detection method according to the embodiment. [Figure 7] FIG. 7 is a control block diagram of the foreign object detection system shown in FIG. [Figure 8] FIG. 8 is a schematic diagram illustrating an overview of tilting or rotating a bottle in an example of an image capturing process in the foreign object detection method according to the embodiment. [Figure 9] FIG. 9 shows an example of image processing parameters set in the foreign object detection method according to the embodiment. [Figure 10] FIG. 10 is a flowchart showing a foreign object detection method according to the embodiment. [Figure 11] FIG. 11 is a flowchart showing the process of the foreign object detection method according to the embodiment. [Figure 12] FIG. 12 is a flowchart showing a frame capture process of the foreign object detection method according to the embodiment. [Figure 13] FIG. 13 is a schematic diagram showing an example of a plurality of regions set in an image captured by each trigger signal generated in the imaging process in the foreign object detection method according to the embodiment. [Figure 14] FIG. 14 is a schematic diagram showing an example of the height of the inspection level for an image captured at a time interval obtained by dividing the time period belonging to the first trigger signal into a plurality of times. [Figure 15] FIG. 15 is a flowchart showing image processing in the foreign object detection method according to the embodiment. [Figure 16] FIG. 16 is a flowchart showing a foreign object detection process of the foreign object detection method according to the embodiment. [Figure 17] FIG. 17 is a schematic diagram showing an example of division of an inspection area for an image captured in a foreign object detection method according to another embodiment. [Figure 18] FIG. 18 is a schematic diagram showing an example of a foreign object detection method according to another embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, a foreign object detection system that can be used in a foreign object detection method according to an embodiment of the present invention will be described with reference to the accompanying drawings.

[0014] [Foreign object detection system configuration] 1 and 2 show a foreign object detection system 10 in an initial (stationary) state. The foreign object detection system 10 has a base 11. The base 11 is preferably fixed to a stable floor or the like. The base 11 supports an outer frame (first frame) 12. In the embodiment, the outer frame 12 is connected to the base 11 so as to be rotatable about a first axis (pivot axis) 13 extending in the vertical direction (the up-down direction in the figure). A mechanism (first rotation mechanism) 14 that rotatably connects the outer frame 12 to the base 11 has a vertical rotation axis (shaft) 15 that extends in the up-down direction along the first axis 13.

[0015] The lower part of the vertical rotation shaft 15 is rotatably supported by a bearing 16 provided on the base 11, and the upper part of the vertical rotation shaft 15 is fixed to the outer frame 12. A large-diameter driven gear 17 is fixed to the bottom surface of the outer frame 12, and the upper end of the vertical rotation shaft 15 is fixed to this driven gear 17.

[0016] The driven gear 17 has external teeth 18 on a circumference centered on the first shaft 13. The driven gear 17 meshes with a small-diameter driving gear 19. The driving gear 19 is fixed to a driving shaft (shaft) 22 that extends in the up-down direction along a vertical axis 21. As shown in the figure, the driving gear 19 has external teeth 20 on a circumference centered on the axis 21, and the external teeth 20 mesh with the external teeth 18 of the driven gear 17. The driving shaft 22 is also drivingly connected to a first rotational driving source 23 fixed to the base 11. The first rotational driving source 23 has an electric motor and a reducer. Preferably, the electric motor of the first rotational driving source 23 is an electric motor that rotates the driving shaft 22 forward and backward.

[0017] According to the first rotation mechanism 14 configured in this manner, the rotation of the drive source 23 is transmitted from the drive shaft 22 via the drive gear 19 and the driven gear 17 to the outer frame 12, thereby rotating the outer frame 12 in the forward and reverse directions (clockwise and counterclockwise directions when viewing the foreign object detection system 10 in Figure 1 from above) around the first shaft 13.

[0018] Outer Frame In the embodiment, the outer frame 12 has a base plate 25 extending in the horizontal direction and a pair of side plates 26 extending upward from both the left and right sides of the base plate 25. As shown in the figure, the pair of side plates 26 are preferably arranged symmetrically about the first axis 13. The left and right side plates 26 are each formed with a through hole 28 centered on an axis (second axis) 27 extending in the horizontal direction near their upper ends. The through hole 28 supports a horizontal rotation axis (shaft) 29 extending along the second axis 27 via a bearing (not shown).

[0019] [Inner frame] The horizontal rotation shaft 29 supports an inner frame (second frame) 30 arranged inside the outer frame 12. In the embodiment, the inner frame 30 has a lower member 31 extending horizontally, a pair of side members 32 extending upward from both the left and right ends of the lower member 31, and an upper member 33 connecting the upper ends of the side members 32 and extending horizontally.

[0020] 1, the lower member 31 and the upper member 33 have holding mechanisms 35 for the container 100 to be inspected. In the embodiment, the foreign object detection system 10 has two holding mechanisms 35 arranged symmetrically on the left and right of the first axis 13. Each holding mechanism 35 has a lower end support portion 37 and an upper end support portion 38 that respectively support the lower end and upper end of the container 100 arranged along a vertical axis 36 that extends in the up-down direction.

[0021] The lower end support portion 37 has a substantially circular recess (not shown) that corresponds to the bottom shape of the container 100. The size of the recess (particularly, the inner diameter) does not need to correspond to a specific container, and may be larger than the maximum bottom outer diameter of the target containers.

[0022] Generally, in the case of a glass wine bottle, the bottom of the bottle has a depression in the center that faces upward, so the bottom end support portion 37 may have a protrusion that corresponds to the depression in the bottle bottom.

[0023] The upper end support portion 38 has a cap 39 that can be placed on the upper end of the container 100 that is closed by, for example, a plug. The cap 39 is supported by a cylinder 40, which is a lifting device attached to the upper member 33 of the inner frame 30. For example, a pneumatic cylinder can be used as the cylinder 40. In this case, the cylinder 40 is fixed to the upper member 33 so that its piston rod advances and retreats along the vertical axis 36, and is configured so that the cap 39 descends based on the driving of the cylinder, thereby holding the upper end of the container 100 arranged along the vertical axis 36.

[0024] [Rotation (tilt) mechanism] The horizontal rotation shaft 29 shown on the right side of FIG. 1 penetrates the side plate 26 of the outer frame 12 and protrudes outward, with a second rotation (tilt) mechanism 42 connected to the protruding portion 41. In this embodiment, the second rotation mechanism 42 has an upper pulley 43 fixed to the protruding portion 41. A belt 44 is looped around the upper pulley 43. The belt 44 is also looped around a lower pulley 45 provided below the upper pulley 43. The lower pulley 45 is fixed to a horizontal drive shaft 47 of a second rotation drive source 46 fixed to the base plate 25 of the outer frame 12. Preferably, the motor of the second rotation drive source 46 is an electric motor that rotates the drive shaft 47 forward and backward.

[0025] According to the second rotation mechanism 42 configured in this manner, the rotation of the second rotation drive source 46 is transmitted from the drive shaft 47 via the lower pulley 45, the belt 44, and the upper pulley 43 to the horizontal rotation shaft 29, thereby rotating the inner frame 30 in the forward and reverse directions (clockwise and counterclockwise in Figure 2) around the second shaft 27.

[0026] [Photographic equipment] 1 and 2, the lower member 31 of the inner frame 30 is provided with a pair of left and right arms 48 that are disposed symmetrically on the left and right sides of the first axis 13 and extend forward (to the left in FIG. 2) from the lower member 31. Each arm 48 supports a camera (digital video camera) 49 that is an imaging device. As shown in the figure, the height of the camera 49 is determined so that the optical axis 50 of the camera 49 intersects with the vertical axis 36 at a right angle as shown in FIG. 2 at approximately the center of the container 100 supported by the inner frame 30 as shown in the figure (in the case of the bottle shown in the figure, near the middle of the body).

[0027] For example, in the case where the container 100 is a bottle 110 having a large-diameter lower body 111, a small-diameter upper neck 112, and a shoulder 113 connecting the body 111 and the neck 112 as shown in FIG. 3, the camera 49 is disposed so that its optical axis 50 is perpendicular to the bottle central axis 101 at approximately the center of the body 111. The distance between the camera 49 and the bottle 110 is preferably determined so that almost the entire body 111 is included within the field of view of the camera 49. Also, for example, in the case where the container 100 is a wine bottle having a front label 121 and a back label 122 attached thereto, the area of ​​the label is minimized when the camera 49 photographs the bottle 110 from a direction perpendicular to the lateral center line 123 connecting the center of the front label 121 and the center of the back label 122 of the bottle 110 (the direction indicated by the arrow in FIG. 3(B)). By photographing the bottle 110 in such a position, it becomes easier to detect foreign objects. Therefore, it is desirable to position the container 100 at such an angle for the bottle 110 to be inspected. In the following description, the container 100 is assumed to be the bottle 110 having the shape shown in the drawings.

[0028] An illumination light source 51 is disposed on the opposite side (rear) of the camera 49 across the bottle 110. As shown in Fig. 2, the illumination light source 51 is fixed to the lower member 31 of the inner frame 30. The illumination light source 51 may be a so-called point light source, but it is preferable to employ a surface light source as shown in the figure.

[0029] The bottle 110 to be inspected by the foreign object detection system 10 is not limited to the type of liquid contained therein. In addition, the bottle 110 and the liquid contained therein do not need to be transparent. For example, wine is generally divided into red wine and white wine. Therefore, the foreign object detection system 10 is required to ensure that the camera 49 captures foreign objects (e.g., cork fragments) that may be mixed into red wine. For this reason, it is preferable that the illumination light source 51 is a light source that emits infrared rays. In addition, it is preferable that the bottle 110 and the liquid filled therein are at least capable of transmitting infrared rays.

[0030] [Basic operation] The basic operation of the foreign object detection system 10 having the above-mentioned configuration and the associated movement of the contents (liquid 102, air bubbles 103, foreign object 104) will now be described.

[0031] First, the outer frame 12 rotates in a forward or reverse direction around a first axis (pivot axis) 13 in a vertical direction based on the driving of the first rotation drive source 23 in the first rotation mechanism 14. In the foreign object detection system 10 of the embodiment, two bottles 110 are disposed symmetrically on the left and right of the first axis 13. Therefore, based on the driving of the first rotation drive source 23, the two bottles 110 are subjected to a pivoting action about the first axis 13.

[0032] Next, based on the driving of the second rotation drive source 46 in the second rotation mechanism 42, the inner frame 30 rotates in the forward and reverse directions (clockwise and counterclockwise in Figure 2) around the horizontal second axis (tilt axis) 27, thereby changing the tilt angle of the two bottles 110.

[0033] For example, as shown in Fig. 4, when the second rotation drive source 46 is driven from an upright state in which the bottle 110 has its central axis 101 oriented vertically (as shown by the solid line; hereinafter, this state will be referred to as the "vertical upright state"), the bottle 110 gradually increases its inclination according to the amount of rotation of the inner frame 30, and passes through a horizontal state in which the central axis 101 of the bottle 110 is oriented almost horizontally (as shown by the dashed-dotted line), and then transitions to a state in which the bottle 110 is upside down. In the figure, the state in which the center line (vector) extending from the center of the bottom of the bottle 110 to the center of the mouth is oriented vertically upward (as shown by the solid line) is referred to as the "vertical upright state", and the state in which the central axis is oriented horizontally (as shown by the dashed-dotted line) is referred to as the "horizontal state". In addition, the state from the vertical upright state in which the central axis is pointed vertically upward to the state in which the central axis is pointed horizontally (the state shown by the dashed dotted line) is called the "upright state," and the state in which the central axis is pointed downward so that the bottom of the bottle is located higher than the horizontal state mentioned above (the state shown by the dashed dotted line and dotted line) is called the "inverted state."

[0034] In the above-described configuration, based on the driving of second rotation drive source 46 in second rotation mechanism 42, bottle 110 can transition from a vertically upright state to an inverted state, for example, as shown in FIG. 4, and conversely, can return from the inverted state to a vertically upright state, as shown in FIG. 5, and the contents (liquid, foreign matter, air bubbles) move inside bottle 110 depending on the state of the bottle at each time.

[0035] For example, as shown in FIG. 6A, when the bottle 110 moves from a vertically upright state to a horizontal state, the air 103 at the mouth of the bottle starts to move toward the bottom along the upper inner surface of the upper neck 112 oriented in the horizontal direction. Also, foreign matter 104 having a specific gravity greater than that of the liquid 102 and accumulated at the bottom of the container starts to move toward the lower inner surface of the body 111 oriented in the horizontal direction under the influence of gravity. Next, when the bottle 110 moves to the inverted state (tilt angle is about 100 degrees) of FIG. 6B, the air bubbles in the upper neck 112 move beyond the shoulder 113 to the body 111 and move upward along the upper inner surface of the body 111. Meanwhile, the foreign matter 104 at the bottom of the body 111 moves down toward the shoulder 113 along the lower inner surface. Next, when the bottle 110 moves to the inverted state (tilt angle is about 160 to 170 degrees) of FIG. 6C, the air bubbles accumulate at the bottom of the body 111, and the foreign matter 104 moves toward the upper neck 112. As shown in the figure, when bottle 110 transitions from a vertically upright state to an inverted state, the movement speed (sinking speed) of foreign matter 104 is smaller than the movement speed (ascending speed) of air bubble 103, and as shown in FIG. 6C, when air bubble 103 has finished moving, foreign matter 104 is still in the moving state.

[0036] The tilting motion of the bottle 110 described above can be performed during the rotation motion of the bottle 110, or can be performed independently of the rotation motion of the bottle 110. However, by combining the tilting motion and rotation motion of the bottle 110, the contents (liquid, gas, foreign matter) in the bottle 110, especially the foreign matter, exhibit a complex movement that combines rising and falling and rotation, which makes it easier to detect the foreign matter.

[0037] [Control Unit] FIG. 7 is a diagram showing the configuration of the control system of the foreign object detection system 10. As shown in the figure, the foreign object detection system 10 has a control unit 55. The control unit 55 is, for example, a computer. The control unit 55 has a CPU 60, which is a general-purpose processor that realizes a predetermined function by executing a program. The CPU 60 is connected to the first rotation drive source 23, the second rotation drive source 46, the cylinder 40, the camera 49, and the illumination light source 51 described above by wire or wirelessly. The control unit 55 and each component included in the control unit 55 realize various processes in the foreign object detection system 10 by calling and executing a control program stored in, for example, a storage unit 63 described later. The control unit 55 and each component included in the control unit 55 are not limited to those that realize a predetermined function by cooperation between hardware and software, and may be a hardware circuit designed specifically to realize a predetermined function. In other words, the control unit 55 and each component included in the control unit 55 can be realized by various processors such as a CPU, an MPU, a GPU, an FPGA, a DSP, and an ASIC.

[0038] The control unit 55 of the foreign object detection system 10 has an input unit 61 which is a typical computer input device such as a keyboard or mouse for inputting necessary information, an operation unit 62 which is equipped with a drive start switch (not shown) or buttons for inputting other necessary signals, a memory unit (RAM / ROM) 63 which stores programs, information, and information input from the input unit 61 required to execute the processes described below, a process processing unit 64 which executes multiple processes (such as rotating the bottle 110) described below to acquire multiple images (frames) required for foreign object detection, an image processing unit 65 which applies necessary processing (e.g., filter processing) to the acquired multiple images (frames) to create data (binary data) required for foreign object determination, a foreign object detection unit 66 which detects foreign objects from the image-processed data, an output unit 67 which is, for example, a display capable of displaying images captured by a camera, processed images, etc., and a drive unit 68 which controls the operation (e.g., rotation direction or rotation speed) of the first rotation drive source 23 and the second rotation drive source 46.

[0039] [Foreign object detection method] The foreign object detection by the foreign object detection system having the above configuration will now be described. For ease of understanding, in the foreign object detection described below, the tilting operation of the bottle 110 includes four processes. The four processes are, for example, as shown in FIG. 8, a first tilting process 1000 in which the bottle 110 in a vertical upright state (the state shown in FIG. 8(a)) is rotated counterclockwise around the central axis 27 by an angle θ1 to become a first inverted state 1001 (the state shown in FIG. 8(b)); a second tilting process 2000 in which the bottle 110 in the first inverted state is further rotated counterclockwise around the central axis 27 by an angle θ2 to become a second inverted state 2001 (the state shown in FIG. 8(b)); a third tilting process 3000 in which the bottle 110 in the second inverted state is rotated clockwise by an angle θ3 to become an upright state 3001 (the state shown in FIG. 8(c)); and a fourth tilting process 4000 in which the bottle 110 in the third inverted state is further rotated clockwise by an angle θ4 to become a vertical upright state 4001 (the state shown in FIG. 8(d)).

[0040] The tilting processes will be mainly described below, but the tilting action in each tilting process can be combined with the pivoting action of the bottle 110 or can be performed alone.

[0041] [Processing] Corresponding to these four processes, in foreign object detection system 10, when a power switch (not shown) is pressed (turned on), CPU 60 initializes information (parameters) required to execute multiple processes (S1 in FIG. 10). The initially set parameters include, in addition to the various information shown in FIG. 9 (described later), the initial set process number N, whether or not images (frames) are acquired during execution of each process, the number n of images acquired during or immediately after the end of each process, or both, various timer values ​​T1 to T3, and flag F.

[0042] When a signal to start a foreign object detection process is input through input unit 61 in this state, CPU 60 executes the processes shown in FIGS.

[0043] Specifically, when a start signal is input (S2), the CPU 60 sequentially executes processing (S3) and frame capture processing (S4) by the process processing unit 64, image processing (S5) by the image processing unit 65, and foreign object detection processing (S6) by the foreign object detection unit 66.

[0044] Referring to FIG. 11, in the process S1, the CPU 60 checks the process number N (S101). In the initial setting (S1), the process number N is set to "1". Therefore, the CPU 60 reads out information corresponding to the process number 1 (first tilt process) from the storage unit 63 (S102). The read out information includes a timer value T1 that manages the drive time of the second rotation drive source 46 and a rotation direction of the motor in the second rotation drive source 46. Information regarding the rotation speed of the motor may also be included. Next, the CPU 60 starts the timer t1 (S103) and starts the motor 46 (S104). This starts the first tilt process described above, and the bottle 110 transitions from the vertically upright state to the first tilt state. After the first tilt process starts, the CPU 60 determines whether or not the flag F (the "trigger signal" of the frame capture process) is set to "0" and whether or not to perform the frame capture process during the execution of the tilt process (S105). The flag F is set to "0" in the initial setting (S1). When the flag F is "0" and the frame capture process is performed during the execution of the tilt process, the CPU 60 sets the flag F to "1" (S106). The flag F may be set to "1" after the count value of the timer t1 reaches a predetermined count value. By configuring in this way, the tilt process starts, and the frame capture process described later can be performed after a predetermined time has elapsed. The CPU 60 drives the motor in the second rotation drive source 46 until the count value of the timer t1 reaches the timer value T1 (S107). When the drive time of the motor in the second rotation drive source 46 reaches the predetermined time (when the count value of the timer t1 reaches the timer value T1), the bottle 110 reaches the first tilt state shown in FIG. 8(b).

[0045] Next, the CPU 60 stops driving the motor in the second rotational drive source 46 (S108), and determines whether or not the flag F is set to "1" (S109). If the flag F is not "1", the CPU 60 sets the flag F to "1" (S110).

[0046] Next, the CPU 60 determines whether the process number N is "4" (S113). For example, immediately after the first tilt process is executed, the process number N is "1". In this case, the CPU 60 increments the process number N (N+1) and returns to the initial process (S101) (S114). On the other hand, if the process number N is 4, that is, the fourth tilt process is completed, the process is terminated.

[0047] In the above explanation, the next tilt process is started immediately after the motor is stopped, but, for example, after the motor is stopped, timer t2 may be started (S111), and the next tilt process may be started when the count value reaches timer value T2 (S112).

[0048] The above processing is executed for each of the first to fourth tilt processes. However, in the first to fourth processes, the combination of the rotary drive source to be driven, and the drive time, rotation direction, and rotation speed of the motor of the rotary drive source are different, and the information is read from the storage unit 63 in process information acquisition (S102), and the rotary drive source to be driven, and the drive time, rotation direction, and rotation speed of the motor of the rotary drive source are controlled based on the read information. The first rotary drive source 23 and the second rotary drive source 46 may be driven simultaneously.

[0049] [Frame capture process] The frame capture process is performed simultaneously with the above-mentioned process or after a predetermined time has elapsed. The predetermined time can be set, for example, by the timer t2 as described above. As shown in FIG. 12, in the frame capture process, it is determined whether or not the flag F is set to "1" (S120). As described above, the flag F is set to "1" during or at the end of each tilt process. If the flag F is "1", the CPU 60 captures a frame of an image captured by the camera 49 (S121). Next, the CPU 60 increments the count value of the counter n1 that counts the number of captured frames (S122). The CPU 60 then determines whether or not the incremented count value of the counter n1 has reached a predetermined number n (the number of frames to be acquired) (S123), and if the count value is less than n, the timer t3 is started (S124) and waits until the count value reaches a target timer value T3 (S125). When the count value of the timer t3 reaches the timer value T3, frames are captured again (S121) In this manner, for one process, n frames are acquired at a fixed interval t3.

[0050] When n frames have been acquired, the CPU 60 resets the counter n1 and the timer t3 (S126), resets the flag F to "0" (S127), and ends the frame acquisition process.

[0051] [Image processing] Next, the filtering process performed in the image processing will be described with reference to FIG.

[0052] 13, in the image processing of the embodiment, regions of interest of different sizes are set in different locations in an image captured by camera 49 according to the attitude of bottle 110, and the region of interest is divided into a plurality of inspection regions of different sizes (four in the embodiment), and filters of different strengths (for example, levels of noise removal capability) are set for each inspection region. Each image processing process described below is executed after the completion of each tilting process, but if a frame is acquired during the execution of a tilting process, the image processing process may be started during the execution of the tilting process.

[0053] [Setting the region of interest / inspection area] For example, in the first image processing process performed after the first tilting process is completed, bottle 110 has just transitioned from an upright state to an inverted state, so most of the air bubbles have already migrated to the vicinity of the bottom of bottle 110, but the foreign matter that was at the bottom is gradually settling from the bottom toward the mouth near the middle of the inverted bottle body while encountering resistance from the liquid. Therefore, in the first image processing process ("filter processing" in Figure 15 described later), the region of interest 1010 is set from near the middle of the body to near the neck, and the region of interest 1010 is divided into four inspection regions 1011, 1012, 1013, and 1014 (first inspection region 1011, second inspection region 1012, third inspection region 1013, and fourth inspection region 1014). The width of inspection region 1014 near the bottom shown in Figure 13(a), where foreign matter is most likely to exist, (the width in the direction of the central axis of the bottle) is set to be the largest, and the width of inspection region 1011 near the neck shown in the same figure, where foreign matter is less likely to exist, is set to be the smallest, and the width gradually decreases from inspection region 1014 to inspection region 1011.

[0054] Next, in the second image processing process performed after the second tilting process, the inversion angle of the bottle 110 becomes even larger, and at this point, most of the air bubbles have migrated to the vicinity of the bottom of the bottle 110, and most of the foreign matter is present near the neck of the inverted bottle. Therefore, in the second image processing process ("filter processing" in FIG. 15 described later), the region of interest 2010 is set wider toward the bottom than in the first image processing process, and the region of interest 2010 is divided into four inspection regions 2011, 2012, 2013, and 2014, with the width of region 2014 close to the bottom, which appears in the figure and where foreign matter is unlikely to exist, being set to the smallest, and the width of the regions is gradually reduced from region 2011 to region 2014.

[0055] Next, in the third image processing process performed after the third tilting process is completed, the bottle 110 is inverted from the inverted state to the upright state, and at this point, most of the air bubbles are present near the neck of the bottle 110. Meanwhile, the foreign matter that was near the neck of the bottle 110 is settling around the middle of the body of the upright bottle while being subjected to the resistance of the liquid. Therefore, in the third image processing process ("filter processing" in Figure 15 described later), the region of interest 3010 is moved toward the bottom compared to the first and second image processing processes, and the region of interest 3010 is divided into four inspection regions 3011, 3012, 3013, and 3014. The width of inspection region 3012 near the middle of the body shown in Figure 13(c), where foreign matter is likely to exist, is set to the largest, while the widths of inspection regions 3011 and 3014 close to the neck and bottom shown in the same figure, where foreign matter is unlikely to exist, are set to the smallest, and the width of inspection region 3013 halfway between inspection regions 3012 and 3014 is set to be larger than inspection regions 3011 and 3014.

[0056] Next, in the fourth image processing process performed after the fourth tilting process, the bottle 110 is shifted to a vertically upright position, and at this point, most of the air bubbles have passed over the neck of the bottle 110 and are present near the upper end mouth. Meanwhile, the foreign matter is sinking in the area around the middle part of the body or on the bottom side thereof while still receiving resistance from the liquid. Therefore, in the fourth image processing process ("filter processing" in FIG. 15 to be described later), the region of interest 4010 is expanded closer to the neck than in the third image processing process, and the region of interest 4010 is divided into four inspection regions 4011, 4012, 4013, and 4014, and the width of the inspection region 4013 slightly below the middle part of the body shown in FIG. 13(d) where foreign matter is likely to exist is set to the largest, the width of the inspection region 4014 below the inspection region 4013 is set to the next largest, and the widths of the inspection regions 4011 and 4012 close to the neck and bottom shown in the same figure where foreign matter is unlikely to exist are set to be small.

[0057] As shown in Fig. 14, the setting of the region of interest and the inspection region can be set multiple times according to the elapsed time even in the same image processing process. As described above, in the first image processing process, since the bottle 110 has just been shifted from an upright state to an inverted state, most of the air bubbles have already moved to the vicinity of the bottom of the bottle 110, but the foreign matter at the bottom is gradually settling from the bottom to the mouth near the middle of the body of the inverted bottle while receiving resistance from the liquid. Therefore, in the first time, which is a predetermined period immediately after the start of the first image processing process, the region of interest 1010 is set from the vicinity of the middle of the body to the vicinity of the neck, and the region of interest 1010 is divided into four inspection regions 1011, 1012, 1013, and 1014, as shown in Fig. 14(a).

[0058] At the second time, which is a predetermined period after the first time has elapsed, the air bubbles have migrated further toward the bottom of the bottle 110 compared to the first time. Therefore, as shown in FIG. 14(b), the width of the region of interest 1020 is set larger toward the bottom side compared to the region of interest 1010. In addition, inspection regions 1021, 1022, 1023, and 1024 are set by dividing the region of interest 1020. At the third time, which is a predetermined period after the second time has elapsed, the air bubbles have migrated further toward the bottom of the bottle 110 compared to the second time. Therefore, as shown in FIG. 14(c), the width of the region of interest 1030 is set larger toward the bottom side compared to the region of interest 1020. In addition, inspection regions 1031, 1032, 1033, and 1034 are set by dividing the region of interest 1030. In this way, even with the same image processing process, by setting multiple time periods based on the timing at which the frames were captured and adjusting the size of the area of ​​interest and inspection area for each time period, the size of the inspection area and its filter strength can be set more effectively in accordance with the distribution of the contents within bottle 110, thereby improving the accuracy of foreign object detection.

[0059] An example of information on the above-mentioned region of interest and inspection region is shown in FIG. 9. In the figure, "first process" to "fourth process" correspond to the first to fourth image processing processes, respectively. FIG. 9 displays parameters set for the first process, which is image processing for a frame acquired in the first frame capture process. "No. 1" indicates a setting item for the first time of the process. "Time" indicates a predetermined period (e.g., the first time period) for which the setting item is applied. In this embodiment, the first time means image processing performed on a frame acquired during a period from the start of frame capture (0 ms) in the first frame capture process until 200 ms has elapsed. The "start coordinate" of the "region of interest" is the coordinate (pixel number) of the upper left corner of the region of interest included in the camera image, the "region of interest width" is the number of pixels in the horizontal direction (x direction) included in the region of interest, and the "vertical start point" and "vertical end point" are the pixels in the vertical direction (y direction) of each inspection region within the region of interest. The "first area" to the "fourth area" respectively represent the inspection areas described above, and are the first inspection area 1011 to the fourth inspection area 1014 in "No. 1" of the first process, for example.

[0060] [Filter Strength] The strength of the filter is set separately for each inspection area included in each of the first to fourth tilting processes. Most of the noise is air bubbles generated by rotating (tilting, swiveling) the container. The strength of the filter (level of noise removal ability) is the level of ability to smooth an image, for example, in the case of a smoothing filter. Therefore, when a strong filter is set, the image is smoothed more and many air bubbles are removed. When noise is removed by such a strong filter, if a foreign object exists in the area to which the filter is applied, the foreign object is likely to be removed by image processing in the same way as the noise. Therefore, the area to which a strong filter is set has a low ability to inspect for foreign objects (hereinafter, appropriately referred to as "inspection level").

[0061] For example, in the first and second image processing processes performed after the first and second tilting processes are completed, since the bottle 110 is in an inverted state, most of the air bubbles are present near the bottom of the bottle 110, and the foreign matter at the bottom settles from near the middle of the body toward the neck while being subjected to the resistance of the liquid. Therefore, in the first and second image processing processes, the filter strength is set low (i.e., the inspection level is set high) for the inspection area near the neck shown in Figures 13(a) and (b), where there is a low possibility of air bubbles and a high possibility of foreign matter, and the filter strength is set high (i.e., the inspection level is set low) for the inspection area near the bottom of the body shown in the same figures, where there is a high possibility of air bubbles and a low possibility of foreign matter.

[0062] Conversely, in the third and fourth image processing processes performed after the third and fourth tilting processes are completed, the bottle 110 is in an upright state, and at this point, most of the air bubbles are present near the neck of the bottle 110. Meanwhile, the foreign matter is experiencing the resistance of the liquid and is settling from around the middle of the body of the upright bottle to the area below it. Therefore, in the third and fourth image processing processes, the filter strength is set low (i.e., the inspection level is set high) for the inspection area from the middle to the bottom of the body shown in Figures 13(c) and (d), where there is a low possibility of air bubbles and a high possibility of foreign matter, and the filter strength is set high (i.e., the inspection level is set low) for the inspection area close to the neck shown in the same figures, where there is a high possibility of air bubbles and a low possibility of foreign matter.

[0063] In the embodiment, the strength of the filter is determined by three elements, namely, "median", "opening" and "lower area limit". "Median" refers to a median filter. "Opening" refers to opening processing. "Lower area limit" is the lower limit of the area of ​​a pixel considered to be a foreign object in a difference image obtained by a difference processing described later. The numerical value shown for each element is related to the noise removal ability of the filter, and the larger the numerical value, the greater the noise removal ability.

[0064] For example, in No. 1 of the table in FIG. 9, the median filters are set to "m6", "m7", "m8", and "m9" for the first inspection area, the second inspection area, the third inspection area, and the fourth inspection area, respectively, and in the first time of the first image processing process shown in FIG. 14(a), a median filter with a larger noise removal capability is set for the lower inspection area where bubbles are likely to exist. The same applies to the opening and the area lower limit. Also, in No. 2 of the table in FIG. 9, the median filters are set to "m5", "m6", "m8", and "m8" for the first inspection area, the second inspection area, the third inspection area, and the fourth inspection area, respectively, and in this embodiment, the filter strength is set smaller (the inspection level is higher) for the second time than for the first time.

[0065] Next, the specific contents of the image processing will be described with reference to Fig. 15. When the frame capture process is completed, the foreign object detection system 10 performs image processing on the frame. Furthermore, when a frame is acquired while the tilt process is being executed, the foreign object detection system may be configured to start image processing while the frame capture process is being executed.

[0066] When the image processing unit 65 confirms that the frame capture process is completed (S201), it acquires processing information for performing image processing (S202). The acquired information includes each parameter shown in FIG. 9, the process in which the frame was captured and the timing of capture, and the number of frames captured in each process. Next, the image processing unit 65 reads out the frame from the storage unit 63 (S203). Then, the image processing unit 65 performs a filter process on the read out frame based on the acquired processing information (S204). In the filter process, an image process that is preferably performed before a difference process described later is performed. Such image processing includes, for example, noise removal by smoothing. In the filter process, the filter process for the process is performed with a filter strength set for the inspection area set as described above according to the acquired process.

[0067] Next, the image processing unit 65 performs difference processing between frames (S205). In the difference processing, a difference is made between the captured frames to create a difference image. The difference between the frames is obtained, for example, by subtracting the other frames from the frame captured at the end of all processes. Also, difference processing may be performed between other frames, such as subtracting the other frames captured during each process from the frame captured at the end of the process. The difference can be performed, for example, using a difference filter that compares a reference image with an input image and outputs the difference between them as an absolute value. By comparing whether the absolute value of the value obtained by the difference is greater than a predetermined threshold value, if an object floating or moving in the bottle 110 exists, a binary image (binary data) in which the object image portion is extracted can be created. The method of creating the binary data may be any of a frame difference method, a background difference method, a codebook method, and the like.

[0068] Next, the image processing unit 65 performs an opening process (S206) to remove minute noise. After performing image processing on the difference image, the obtained binary data is stored in the storage unit 63 (S207) and displayed on the output unit 67 (S208), and the image processing ends.

[0069] [Foreign object detection processing] Next, in the foreign object detection method, foreign object detection processing is performed on the obtained binary data by the foreign object detection unit 66. The specific contents of the foreign object detection processing will be described with reference to FIG.

[0070] First, the foreign object detection unit 66 reads the binary data from the storage unit 63 (S301). The foreign object detection unit 66 calculates the size of the object image portion (the area of ​​pixels) from the binary data, and judges an object whose size exceeds a predetermined threshold to be a foreign object (S302). As a result, even if an object image of a very small size is obtained by subtraction, the foreign object detection unit 66 can judge an object image whose size is equal to or smaller than the predetermined threshold to be noise such as a very small air bubble. The difference portion judged to be a very small noise is deleted from the binary data.

[0071] If the foreign object detection unit 66 determines that a foreign object is present through the above image processing, it issues a warning that a foreign object is present (S303), for example by displaying on the output unit 67 that a foreign object is present, or by highlighting and displaying pixels determined to be foreign objects. If it determines that no foreign object is present, or after issuing a warning, the foreign object detection process ends.

[0072] The above is a foreign object detection method according to an embodiment of the present invention using a foreign object detection system. According to this foreign object detection method, the noise removal process in the image processing can be set according to the elapsed time from the start of tilting and according to the region of the container. Therefore, foreign objects present in the container can be more efficiently detected by performing foreign object inspection while dynamically changing the noise removal process according to the distribution of contents in the container that changes as the container rotates.

[0073] 9 are initially set at the start of the process in the above description, but the present invention is not limited to this, and for example, after the frame capture process is completed, the acquired frames may be displayed on the output unit 67, and parameters may be set while checking each frame. By setting parameters while checking the frames in this way, an appropriate filter can be selected for a more appropriate area, and noise can be effectively removed.

[0074] In addition, the parameter settings shown in FIG. 9 change the setting of the inspection area and the setting of the filter strength applied to the inspection area for the frame to be image-processed in each image processing process according to the time when the frame was captured. However, this is not limited to this, and for example, the same setting may be used in the same image processing process. That is, the same multiple inspection areas may be set for a frame to be image-processed in a certain image processing process regardless of the time when the frame was captured, and the filter strength for the inspection area may be set. In addition, the noise removal process is performed before and after the creation of the difference image, but there may be an inspection area that is performed only in one of the two cases, or there may be an inspection area that is not performed at any timing.

[0075] In the above description, the foreign object detection method performs foreign object detection using four processes, but the method is not limited to this. For example, foreign object detection may be performed using two processes, which involve rotating and photographing the object so that it is in an inverted state, and then rotating and photographing the object so that it is in an upright state. Foreign object detection may also be performed using a number of processes other than the above.

[0076] In addition, the container is initially placed in a vertical upright state, tilted to an inverted state, and then tilted back to the upright state, but this is not limited to the above. For example, the container may be initially placed in an upright state, tilted to an upright state, and then tilted back to the upright state, and the rotation may be performed, and the image taken during this period may be used for foreign object detection. The container may be turned upside down more times, or may be turned upside down only once. In this way, the container can be tilted or rotated in any manner to detect foreign objects.

[0077] In the foreign object detection method according to the embodiment of the present invention, an image processing method such as a median filter is used as a filter for removing noise such as air bubbles, but this is merely an example and is not limited to this. For example, instead of or in addition to the median filter, other smoothing filters (e.g., averaging filters or Gaussian filters) may be used, and any image processing such as morphological processing other than the opening process may be performed.

[0078] In the above foreign object detection method, the median filter is applied to an image captured before creating a difference image by subtraction, which will be described later, and the opening process and the lower limit area value are applied to the difference image. However, this is not limited to this, and any filter, etc. may be applied at any timing.

[0079] In the foreign object detection method according to the embodiment of the present invention, the bottle 110 is divided into regions in the vertical direction along the central axis of the container, but this is not limited thereto. For example, the bottle 110 may be divided in the horizontal direction perpendicular to the central axis of the bottle 110 and the shooting direction (the direction corresponding to the second axis 27 in FIG. 1). By dividing the bottle 110 in the horizontal direction in addition to the vertical direction, more flexible noise removal parameters can be set for each region. Hereinafter, the regions are referred to as small regions. The small regions may be set to be symmetrical or asymmetrical with respect to the central axis.

[0080] For example, in a container such as a wine bottle, air bubbles are likely to appear as noise at the boundary areas such as the horizontal ends due to the mirror effect. Therefore, when dividing the image only vertically to set the regions, it may be necessary to set a noise removal filter that does not cause noise even if air bubbles are present at the boundary areas. If such a setting is used, the inspection level for foreign objects will be reduced.

[0081] On the other hand, if the inspection area can be divided both vertically and horizontally, a low inspection level noise removal filter can be set for the boundary area so that air bubbles can be removed even if they are reflected large due to the mirror effect, and a high inspection level noise removal filter can be set for the center area other than the boundary area. Figure 17 is an example of a schematic diagram of the inspection area. In Figure 17, in order to improve visibility, the bottle 110 is shown by a dashed line, and the divided inspection area (for example, a small area 2021 in the center and a small area 2022 in the boundary area) is shown by a solid line.

[0082] Furthermore, if the inspection area can be divided in the horizontal direction, the photographing process for detecting foreign objects can also be performed in a direction along the second axis 27, which is different from that of the foreign object detection system 10 according to the embodiment of the present invention. In the foreign object detection system 10, the camera 49 preferably photographs in the same direction as the rotational direction (or the opposite direction) that tilts the bottle 110 (for example, in a direction along the optical axis 50 perpendicular to the second axis 27 and the vertical axis 36). However, if the inspection area can be divided in the horizontal direction, the camera can be oriented to photograph the bottle 110 with its optical axis facing in the same direction as the second axis 27, which is the rotational axis that tilts the bottle 110. This makes it possible to obtain an image of the bottle 110 photographed from the side, for example, as shown in FIG. 18, and to photograph the movement of air bubbles along the upper side of the bottle 110 to the shoulder 113 of the bottle 110 from the side. This allows the detection level of small areas located above, such as small area 2022 near shoulder 113, to be set low, and the detection level of other small areas to be set high, thereby making it possible to more accurately remove noise and detect foreign objects.

[0083] In the above description, the multiple inspection areas are arranged continuously in the vertical direction without any gaps between them, but the multiple inspection areas may be arranged at intervals in the vertical direction. For example, the inspection areas may be provided near the shoulder 113 and near the bottom of the bottle 110 with a gap between each of the inspection areas.

[0084] In the foreign object detection method according to the present embodiment, the container is tilted to move the foreign object in the container, and the movement is continuously photographed to detect the foreign object. In addition, in order to remove elements other than the foreign object caused by the tilt, a filter having a noise removal capability is set in a plurality of inspection areas set in the container, and the noise removal capability is changed according to the position of the container, thereby performing appropriate noise removal. In addition, since the position, distribution, size, and other conditions of the foreign object and noise are biased depending on the timing of the photographing, multiple time periods are set, and different noise removal capabilities are set according to the time periods, thereby performing noise removal using an appropriate noise removal capability according to the expected conditions of the foreign object and noise. Therefore, it is not necessarily necessary to tilt the container as shown in the foreign object detection method according to the present embodiment, and the noise removal capability may be changed according to the size and position of the inspection area depending on the conditions of the foreign object to be detected and the position, distribution, and size of the element that becomes the noise to be removed, and the noise removal capability may be changed according to the time period in addition or alternatively to detect the foreign object. For example, instead of tilting the container as shown in the foreign object detection method according to the present embodiment, the foreign object in the container may be moved by vibrating the container. Alternatively, a rotation axis may be set at an arbitrary position where centrifugal force can be applied to the container by rotating the container, and the container may be rotated while being kept vertical or while being tilted, thereby moving the foreign object in the container. Instead of the tilting process in the foreign object detection method according to the present disclosure, any motion as described above may be applied to the container to move the foreign object in the container, and the movement may be detected to detect the foreign object in the container.

[0085] The foreign object detection method and the foreign object detection system described in all claims of the present disclosure are realized by cooperation with hardware resources, such as a processor, a memory, and a program. [Industrial Applicability]

[0086] As described above, according to the invention disclosed herein, a foreign object detection method can be provided that reduces the rate of false positives by photographing a container containing liquid, setting multiple regions in the photographed image of the container, and efficiently removing noise by changing the noise removal capability of the filter to be applied depending on the part of the container and the time of day the container was photographed, and thus can be suitably used in this type of field in which foreign objects in liquid contained in a container are detected. [Explanation of symbols]

[0087] 10 Foreign object detection system 49 Camera 64 Process Processing Section 65 Image Processing Unit 66 Foreign object detection unit 100 containers 110 Bottles 1010 Area of ​​Interest 1011 First inspection area 1012 Second inspection area 1013 Third Inspection Area 1014 4th Inspection Area

Claims

1. An imaging process for imaging a container made of a light-transmitting material containing a liquid; an image processing process for filtering the captured image of the container; a detection process for detecting foreign objects contained within the container based on the filtered image; The method is characterized in that the image processing process includes setting a plurality of inspection areas in the image of the container, setting a plurality of time periods based on the timing at which the image of the container was taken, and setting a filter having noise removal capability in the plurality of inspection areas for each of the plurality of time periods.

2. The method according to claim 1 , wherein the plurality of inspection regions are set consecutively without any space therebetween or set intermittently with spaces therebetween along a central axis of the container.

3. At least one of the plurality of inspection regions is divided into a plurality of small regions along a lateral direction perpendicular to a central axis of the container; The method according to claim 1 , further comprising setting a filter having a noise removal capability for the plurality of small regions.

4. The method according to claim 3 , wherein the plurality of small regions are set symmetrically with respect to a central axis of the container.

5. and a tilting process for tilting the container. The photographing process photographs the container during or after the tilting process, or both; The method according to any one of claims 1 to 4, wherein the image processing process further comprises setting a filter having a noise removal capability in the plurality of inspection areas according to the orientation of the container when the image was captured.

6. 6. The method of claim 5, wherein the tilting process comprises tilting the container by rotating it in a positive or negative direction or in both positive and negative directions about a horizontal axis perpendicular to a central axis of the container.

7. 7. A method according to claim 5 or 6, wherein the image processing process sets the plurality of time periods during and / or after the tilting process.

8. A method for detecting foreign matter contained in a container, the method comprising: photographing the container while turning the container upside down and made of a light-transmitting material in which a liquid is contained; setting a plurality of regions in an image of the container; determining whether a foreign object is present in an image of a container in an upright state by setting a filter having a weaker strength in an area near the bottom of the container than in an area near the mouth of the container; and identifying foreign objects in an image of an inverted container by setting a weaker filter in an area near the mouth of the container than in an area near the bottom of the container.

9. a tilting process for tilting a container made of a light-transmitting material containing a liquid; an imaging process for imaging the container during or after the tilting process, or both; an image processing and foreign object detection process for detecting foreign objects contained in the container based on the captured image of the container; The image processing and foreign object detection process is a method in which a region of interest is set in an image of the container, foreign objects are detected in the set region of interest, and the size of the region of interest is changed in the axial direction depending on the orientation of the container when the image was taken.

10. The method according to claim 9 , wherein the image processing and foreign object detection process comprises dividing the region of interest in the direction of the central axis to set a plurality of inspection regions, and setting filters having different filter strengths for the plurality of inspection regions.

11. The method according to claim 10 , wherein the plurality of inspection regions are set consecutively without any space therebetween or set intermittently with spaces therebetween along a central axis of the container.

12. At least one of the plurality of inspection regions is divided into a plurality of small regions along a lateral direction perpendicular to a central axis of the container; The method according to claim 10 or 11, wherein a filter having a noise removal capability is set for the plurality of small regions.

13. The method according to claim 12 , wherein the plurality of small regions are set symmetrically with respect to a central axis of the container.

14. A means for turning upside down a container made of a light-transmitting material containing a liquid; A means for photographing the container; A means for processing the captured image of the container, The means for processing the image includes: A region setting means for setting a plurality of regions in an image of the container; an image processing means for filtering the image by setting a filter having a noise removal capability for the plurality of regions in accordance with the posture of the photographed container; and a foreign object detection means for detecting foreign objects contained in the liquid based on the filtered image.

15. A means for turning upside down a container made of a light-transmitting material containing a liquid; A means for photographing the container; A means for processing the captured image of the container, The means for processing the image includes: a region setting means for setting a region of interest in an image obtained by photographing the container; a foreign object detection means for detecting a foreign object contained in the liquid in the region of interest, A foreign object detection system, wherein the region setting means changes the size of the region of interest depending on the posture of the photographed container.

16. A means for turning upside down a container made of a light-transmitting material containing a liquid; A means for photographing the container; A means for processing the captured image of the container, The means for processing the image includes: a region setting means for setting a plurality of regions in an image of the container in accordance with the orientation of the container; an image processing means for filtering the image by setting a filter having a noise removal capability for the plurality of regions; and a foreign object detection means for detecting foreign objects contained in the liquid based on the filtered image.

17. an imaging step of imaging a container made of a light-transmitting material containing a liquid; an image processing step of filtering the captured image of the container, the image processing step including: setting a plurality of inspection areas in the image of the container; setting a plurality of time periods based on the timing at which the image of the container was captured; and setting a filter having noise removal capability in the plurality of inspection areas for each of the plurality of time periods; a detection step of detecting foreign objects contained in the container based on the filtered image; A computer program for causing a computer to execute the above.

Images