JPH06331567A - Defective-can inspection apparatus - Google Patents

Abstract

PURPOSE:To obtain a defective-can inspection apparatus which can detect a tab without being affected by the density of a painted color on a can. CONSTITUTION:An image processing device 38 is provided with a virtual-cap processing part 61, and the virtual-cap processing part 61 is constituted of a cap detection part 81, a cap-pattern memory part 82 and a cap creation part 63. A cap is supplied in the cap creation part 83 by detecting a similar pattern from the cap-pattern memory part 82 on the basis of a cap appearing on the surface of a part even when the cap is not imaged in the cap detection part 81 in the virtual-cap processing part 61. A tab inside a processed area can be detected while it is made to appear in a crescent shape inside a dark part by using the supposed cap as a reference.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a defective can inspection device, and more particularly to an improvement of a defective can inspection device for inspecting a defective can by image processing.

[0002]

2. Description of the Related Art Usually, a can is sealed by winding a can lid around a can body. In this winding operation, the upper end of the can body 12 of the can 10 is bent outward as shown in FIG.
It is bent into a U-shape while being inserted in the space of the bent portion 12a, and is crimped onto the can body 12 and wound.

However, when the outer peripheral portion 14a of the can lid 14 is not inserted into the space between the bent portions 12a of the can body 12, as shown in FIG.
The tip of 2a is exposed on the body surface of the can, resulting in a so-called tongue dashi can. Needless to say, in this state, the can is not kept in a good sealed state, the contents of the can leak, or the microorganisms invade the can. Therefore, it has been attempted to eliminate such a tongue can in the manufacturing process.

[0004] As a general method of eliminating the tongue dashi can, the fact that the pressure inside the can is improper due to incomplete can winding is used.
Although a method of inspecting the can internal pressure with a can and the like is well known, a more direct inspection method has been demanded because a change in the can internal pressure does not always occur immediately.

In addition to these demands, in recent years,
Image processing of can side images is considered to enable direct detection of the tongue stick part. The conventional method will be described with reference to FIG. In FIG. 16, the can 10 after the can lid is fastened is installed on a belt conveyer 20 on which the can 10 is conveyed, a ring light 22 as an illuminating device is arranged on the belt conveyer 20, and a tightly fastened part (base) of the can 10 is provided. Four CCD cameras 24, 26, 28, 30 are arranged as concentric circles with the ring light 22 at equal intervals as a photographing device for photographing. Further, below the ring light 22, the outputs of the can intrusion sensors 32, 34, and 36, each of which includes a light emitter and a light receiver sandwiching the belt conveyor 20, are input to a high-speed image processing device 38 as an image processing device, and the inspection device is inspected. If it is determined that the can has entered the inspection site, the image data is taken in from each of the cameras 24, 26, 28, 30 via the camera controller 40 and image processing is performed.

The result of this image processing is displayed on the display 42, and the can 10 which is determined to be a tongue stick can is removed from the belt conveyor 20 by a removing device (not shown).

FIG. 17 is a side view of the main part of the apparatus of this embodiment. As is clear from the figure, the ring light 22 has a diameter larger than the diameter of the can 10 and is arranged above the can 10. The illuminating light 44 is incident from the upper portion of the can lid winding tightening portion (base) 46 at an incident angle θ of about 35 degrees, and excellent tongue sticking detection is performed.

Therefore, as shown in FIG. 19A, when the can lid winding tightening portion (base) 46 is normal, the irradiation lights 44a and 44b are applied to the can lid winding tightening portion (base) 46 (outer surface of the can lid). The irradiation light 44c and 44d are irradiated to the outer wall 12 of the can body.
Is irradiated. The boundary between the can lid winding tightening portion (base) 46 and the can body outer wall 12 is defined by the can lid winding tightening portion (base) 46.
Will be the shadow of.

As a result, by photographing with the cameras 24, 26, 28 and 30 from the side of the can lid winding tightening portion (base) 46, the brightness of one line in the vertical direction becomes as shown in FIG. 19B. The portion (1) in the figure is the can lid winding tightening portion (base) 46.
And (2) is a can lid winding tightening part (base) 4
The dark part which is the shadow of 6, and (3) is the bright part which shows the part of the can barrel 12 where the irradiation light hits.

On the other hand, as shown in FIG. 20A, when the tongue sticking portion 48 is present in the can lid winding tightening portion (base) 46, a photographing result different from that shown in FIG. 19 is obtained. That is, the irradiation light 44a, 44b is applied to the can lid winding portion (base) 46 in the same manner as shown in FIG.
The c and 44d are radiated to the tongue extension 48, and the radiated light 4
4e and 44f are applied to the outer wall 12 of the can body. The boundary between the can lid winding tightening part (base) 46 and the tongue out part 48 and the boundary between the tongue out part 48 and the can body outer wall 12 are dark parts.

As a result, by photographing with the cameras 24, 26, 28, and 30 from diagonally above the outside of the can lid winding tightening portion (base) 46, the brightness of one line in the vertical direction of the tongue sticking portion 48 is shown in FIG. 20B. As shown. In the figure, (1) is a light portion showing the can lid winding tightening portion (base) 46, (2) is a dark portion which is a shadow of the can lid winding tightening portion (base) 46, and (3) is a tongue sticking portion. 48 is a bright portion, (4) is a dark portion showing the shadow of the tongue-out portion 48, and (5) is a bright portion showing the outer wall of the can body 12.

Comparing FIGS. 19B and 20B, FIG.
A peak 50 indicating the edge of the tongue sticking portion 48 is clearly drawn on B, and the tongue sticking can inspection can be performed very easily by detecting this peak.

21 to 23 show actual image processing steps of the tongue out edge detection processing unit 60 which is a main component of the high speed pixel processing device 38. The tongue out edge detection processing unit 60 is composed of an edge enhancement, binarization processing unit 71, a differentiation processing unit 72, a base detection unit 73, and a tongue out detection unit 74.

First, at the time of photographing by the cameras 24, 26, 28 and 30, as shown in the multi-valued image of FIG.
2. The can lid winding tightening part (base) 46, the tongue out part 48, and the can body 12 are shown in the image. In the processing sections Ds to D _E , the edge emphasis and binarization processing unit 71 binarizes the edge emphasis and the brightness and darkness at a constant threshold level, and FIG.
Create image data as shown in. As a result, the tongue projecting portion 48 can be made to appear in a crescent shape in the dark area.

Next, the differential processing section 72 performs a differential processing on the image data line of FIG. 22 on the image scanning line A, and outputs respective light and dark changes as pulses, as shown in FIG.

That is, first, the base detecting unit 73 detects the pulse 100 indicating a light-dark change at the fall of the can lid winding portion (base) 46, and then the fall of the base 46 is used as a reference. The tongue sticking detection unit 74 outputs a pulse 102 indicating a dark-light change at the rising of the tongue sticking portion 48, and a pulse 104 indicating a light-dark change at the trailing edge of the tongue sticking portion 48.
By detecting the width of the tongue, the tongue sticking portion can be detected and the tongue sticking length can be measured.

[0017]

Conventionally, when the coating color of the can body is dark, such as black or brown, there is a bright and dark color indicating the edge of the tongue stick in the dark part of the can winding part (base) and the can body. Depending on the type of can, there are cans with a painted color and a bright color (eg light brown, green, purple) or white (eg white, cream), depending on the type of can. If you illuminate the body of the can with high reflectance, the body of the can will shine and cause noise, so it is necessary to lower the illumination, and the mouthpieces at both ends of the field of view and the tongue to be detected have a mouthpiece that is smaller than the front. Due to the curved shape, it was difficult to see the mouthpiece and the tongue to be detected in some cases.

Although it is possible to project the mouthpiece and the tongue to be detected by making the illumination brighter, in that case, the can body may illuminate due to the reflection of the illumination and cause noise, which may cause erroneous detection. It was Moreover, since the tongue detection processing area is based on the mouthpiece, the tongue stick cannot be detected unless the mouthpiece is reflected. In addition, there is a problem in that the can body part has a high reflectance because of the high reflectance of the can body part, and the same phenomenon as that of the tongue occurs even without the tongue part.

The present invention has been made in view of the problems found in the above-mentioned prior art, and the purpose thereof is to check for defective cans which can maintain excellent detection performance without being affected by the density of the coating color of the cans. To provide a device.

[0020]

A defective can inspection device according to the present invention comprises an illuminating device for irradiating light from obliquely upper outside of a can lid winding portion (base) of a can to be inspected, and the can lid winding portion. An image capturing device that captures an image of the vicinity of the can winding portion (base) from the side, and the image obtained from the image capturing device shows the edge of the tongue in the dark part of the boundary between the can winding portion (base) and the can body. Tongue out edge detection processing unit that detects whether there is a bright part, virtual mouth processing unit that virtually processes the base when the coating color of the can body is colored and bright, and the coating color of the can body is white In this case, a tongue protrusion depression detection processing unit that processes the tongue depression based on the density signal of the multi-valued image is provided.

[0021]

As described above, in the device of the present invention, the illuminating device is arranged obliquely above the outside of the can, and the can lid winding portion (base) is irradiated with light from the outside diagonally above, so that the can lid winding portion ( Mouth) becomes the light part, the boundary between the can lid winding part (base) and the tongue extension part becomes the shadow of the can lid winding part (base) and becomes the dark part, and the tip of the tongue part becomes the light part and the tongue extension part If the border of the tongue and the body of the can is a shadow of the tongue sticking part and becomes a dark part, and if the color of the can is black or brown is dark, the base part and the tongue sticking part are embossed in a crescent shape. The tongue sticking out is detected by image processing.In addition, depending on the type of can, when the color of the can is colored and it is bright or white, the field of view is illuminated by the illumination that does not cause false detection due to reflection from the body of the can. If the base is not visible at both ends, the base is virtualized and the tongue detection process is performed using the virtual base The rear is set to detect the tongue, and when the can color is white, the tongue cannot be detected because the can body and tongue extension have the same density in the binary image, and the density of the multi-valued image cannot be detected. By utilizing the signal drop, the tongue stick can be detected with extremely high sensitivity.

Moreover, since it does not depend only on the spectral sensitivity characteristics of the tongue sticking portion and other portions, excellent detection performance can be maintained regardless of the type of coating color of the can.

[0023]

【Example】

Example 1. FIG. 1 is a view showing an embodiment of the present invention. In FIG. 1, a ring light 22 as an illuminating device is arranged on a belt conveyor 20, and further four as an image taking device for taking an image of a winding tightening portion (base) of a can 10. CCD cameras 24, 26, 28 and 30 are arranged concentrically with the ring light 22 at equal intervals. Further, below the ring light 22, the outputs of the can intrusion sensors 32, 34, and 36, each of which includes a light emitter and a light receiver sandwiching the belt conveyor 20, are input to a high-speed image processing device 38 as an image processing device, and an inspection device is provided. When it is judged that the can has entered the inspection site of
Image data is taken in from each of the cameras 24, 26, 28, 30 via the camera controller 40 and image processing is performed.
The result of the image processing device is displayed on the display 42, and the can 10 which is determined to be a tongue sticking can is removed from the belt conveyor 20 by a removing device (not shown). The above is the same as in the conventional apparatus when the can is painted in a dark color such as black or brown. In FIG. 1, the image processing device 38 has a virtual mouthpiece processing section 61 and a tongue out edge detection processing section 60.
2 and 3 show actual image processing steps.

In the case of a can with a bright and colored coating color (for example, light brown, green, purple), the reflectance of illumination is high,
When photographing with the cameras 24, 26, 28 and 30, as in the conventional apparatus, in the tongue out edge detection processing unit 60 as shown in FIGS. 2A and 2B, this is performed in the processing sections Ds to D _E ,
Edges are emphasized, and the light and dark are binarized by a constant threshold bell to create image data as shown in FIG. FIG. 2A shows a case where the tongue stick is at the center.
B is the case where there is a tongue stick at the end. As shown in FIGS. 2A and 2B, the base 46 is curved at both ends of the field of view as compared with the front, so that the amount of light reaching the camera is small and the base may be dark and the base may not be easily reflected.

Further, as shown in FIG. 2C, it is possible to brighten the end including the end by making the illumination brighter so that the base is projected, but in that case, the center of the can body shines, causing noise and causing a false detection. became. Therefore, in order to reduce noise, the base 46 is not actually shown as shown in FIG. 2D, but as shown in FIG. 3D, a virtual base 46 is made to make the base 46 virtual.

In the tongue sticking detection process, the base 46 is used as a reference to form a processing area, and therefore the sticking out detection process cannot be performed and the tongue cannot be detected unless the base 46 is reflected. Image processing device 3
8 has a virtual mouthpiece processing unit 61, and a virtual mouthpiece processing unit 6
Reference numeral 1 includes a mouthpiece detecting section 81, a mouthpiece pattern memory section 82, and a mouthpiece creating section 83. Virtually, the mouthpiece 46 is created by detecting a similar pattern from the mouthpiece pattern memory 82 from the mouthpiece 46 that is partially shown on the surface even if the mouthpiece detection unit 81 in the virtual mouthpiece processing unit 61 does not show all the mouthpieces. The base 46 is hypothesized in the section 83. It is possible to detect the tongue stick out in the processing area by making it appear as a crescent moon in the dark part based on this virtual base. Base 4
The hypothesis of 6 detects the similar mouthpiece pattern by the omission of both ends,
Not only is it externally inserted, but it may come off in the middle depending on the shape of the canned portion (base) 46, but in this case as well, it is inserted and connected from a similar base pattern.

3A, 3B, 3C, and 3D are obtained by differentiating FIG. 2 with the image scanning line A, and correspond to FIGS. 2A, 2B, 2C, and 2D with respective light-dark change pulses, and in FIG. Although not shown, the base 46 is hypothesized in FIG. 3D. In FIG. 3, FIG. 3A, 3B, 3C, 3D
Detects the tongue protrusion, and in FIG. 3B, the tongue protrusion 48 is not detected because the base 46 is not shown. In FIG. 3C, it is possible to project the base 46 including the end portion when the illumination is brightened, but the central portion of the can body illuminates, and the rising signal of the can body is erroneously detected as the rising signal and the tongue sticking out. Detects the tongue stick 48 even though there is no 48.

Example 2. FIG. 4 is a diagram showing another embodiment of the present invention. In FIG. 4, an image processing device 38 has a tongue-out fall detection processing unit 62. The tongue drip detection processing unit 62 includes a base detection unit 91, a multi-valued image processing unit 92,
The tongue stick detection unit 93 is configured. 5 and 6 show actual image processing steps.

FIG. 5 shows that when the coating color of the can is white (white or cream), the edges are emphasized in the processing sections Ds to d _E as in the conventional apparatus when the images are taken by the cameras 24, 26, 28 and 30. Light and dark at a certain threshold level 2
6A and 6B are the tongue image 2 of the image data that has been digitized.
Since the value image has the same density in both the normal part and the tongue part, FIG.
Even if the differential processing is performed, a signal similar to the rising of the tongue is generated even in the normal part, which causes erroneous detection. Therefore, the mouthpiece detection unit 91 of the tongue sticking-out detection processing unit 62 detects the mouthpiece, sets the processing area with the mouthpiece 46 as a reference, and the multivalued image processing unit 92 sets the tongue multivalued as shown in FIG. 6C. It is detected that the density signal of the image becomes a dark portion due to the difference between the thickness portion of the tongue and the can body portion, and the signal falls, and the tongue sticking detection unit 93 detects the tongue sticking portion 48. As shown in FIG. 6D, the density signal of the normal multi-valued image of the can body portion does not fall. Therefore,
The tongue can be detected by detecting the drop of the tongue signal using the density signal of the multivalued image.

Example 3. FIG. 7 is a diagram showing another embodiment of the present invention. In FIG. 7, the image processing device 38 has a virtual mouthpiece processing section 61 and a tongue drop detection processing section 62, and the coating density of the tongue stick can is colored. If it is bright, the base 46 is not actually shown in the acceleration base processing unit 61 in order to reduce noise, but a virtual base 46 is created to make the base 46 virtual. If the color of the tongue dashi can is colored and bright,
FIG. 8 is image data obtained by binarizing light and dark in the processing sections Ds to D _E at a constant threshold level.
A and 9B are the differential processing of the binary image of FIG. 8, and FIGS. 9C and 9D are the density signals of the multi-valued image of FIG. 8, and the virtual of the mouthpiece is similar to the mouthpiece that partially appears on the surface in the virtual mouthpiece processing unit 61. A base pattern is selected and extrapolated or interpolated to virtualize the base. The virtual mouthpiece processing unit 61 is the same as that of the first embodiment, and the tongue protrusion drop detection processing 62 is the same as that of the second embodiment.

Example 4. FIG. 10 is a diagram showing another embodiment of the present invention. In FIG. 10, the tongue sticking edge detection processing unit 60 and the tongue sticking out detection processing unit 6 are provided in the image processing device 38 in FIG.
If the coating color is black or the density of brown is high, the tongue sticking out detection processing unit 60 determines whether or not there is a bright portion indicating the edge of the boundary between the can lid winding tightening portion (base) 46 and the can body portion. Is detected. Further, when the coating color of the can is white, FIG. 11 shows that the lightness and darkness are 2 at a constant threshold level in the processing sections D _{S to} D _E.
12A and 12B are the image data subjected to the binarization processing, FIGS. 12A and 12B are the differential processing of the binary image of FIG. 11, and FIGS.
Tongue drop detection processing unit 62 based on the density signal of the multivalued image of
Detects tongue sticking out by performing image processing on the density signal of a multi-valued image. The tongue sticking edge detection processing unit 60 is the same as the conventional one, and the tongue sticking out fall detection processing 62 is the same as the twenty-first embodiment.
Is the same as.

Example 5. FIG. 13 is a view showing another embodiment of the present invention. In FIG. 13, a virtual mouthpiece processing section 61, a tongue out edge detection processing section 60, and a tongue out depression detection processing section 62 are provided in the image processing apparatus for coating. When the color black or brown has a high density, the tongue sticking out detection processing unit 60 detects whether or not there is a bright portion indicating the edge of the boundary between the can lid winding tightening portion (base) 46 and the can body portion. Further, when the coating color of the can is colored and bright, FIG. 14 is image data obtained by binarizing the lightness and darkness at a constant threshold level in the processing sections D _{S to} D _E , and FIGS. 15A and 15B are the binary images of FIG. Differentiation of,
15C and 15D are density signals of a multi-valued image, and when the base 46 is not reflected, the virtual base processing unit 61 virtualizes the base 46. Further, when the coating color of the can is white, the tongue drip detection processing unit 62 detects the tongue by image-processing the density signal of the multivalued image. The virtual base processing unit 6
1 is the first embodiment. The tongue sticking edge detection processing unit 60 is the same as that of the conventional device, and the tongue sticking out detection processing unit 62 is the same as that of the second embodiment.

As described above, according to the apparatus of the present invention, the irradiation light is applied from diagonally above and outside of the can lid winding tightening portion (base) 46, and the tongue sticking edge detection processing unit 60 detects the edge of the tongue sticking portion. The virtual base processing unit 61 detects the virtual base 46, and the tongue drop detection processing unit 62 image-processes the density signal of the multi-valued image to make the base 46 and the tongue a bright can with a paint color or a white system. Even for a can in which the projecting portion 48 is difficult to be reflected, an accurate inspection of the tongue-out portion 48 can be performed regardless of the coating color density of the can or the like.

[0034]

As described above, according to the apparatus of the present invention, the irradiation light is applied obliquely above the can lid winding tightening portion (base) to detect the edge of the tongue sticking portion, and the color of the coating color can be detected. Even if the base or tongue is difficult to see with a bright can or white can, a virtual base is created, and the tongue is detected by the drop of the signal of the multi-valued image, so that the density of the paint color of the can is concerned. Therefore, it is possible to perform an accurate tongue buttock inspection.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of an image processing process according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram of an image processing process according to a second embodiment of the present invention.

FIG. 4 is a diagram showing a schematic configuration of a second embodiment of the present invention.

FIG. 5 is an explanatory diagram of an image processing process according to the second embodiment of the present invention.

FIG. 6 is an explanatory diagram of an image processing process according to the second embodiment of the present invention.

FIG. 7 is a diagram showing a schematic configuration of a third embodiment of the present invention.

FIG. 8 is an explanatory diagram of an image processing process according to a third embodiment of the present invention.

FIG. 9 is an explanatory diagram of an image processing process according to the third embodiment of the present invention.

FIG. 10 is a diagram showing a schematic configuration of a fourth embodiment of the present invention.

FIG. 11 is an explanatory diagram of an image processing process according to the fourth embodiment of the present invention.

FIG. 12 is an explanatory diagram of an image processing process according to a fourth embodiment of the present invention.

FIG. 13 is a diagram showing a schematic configuration of a fifth embodiment of the present invention.

FIG. 14 is an explanatory diagram of an image processing process according to a fifth embodiment of the present invention.

FIG. 15 is an explanatory diagram of an image processing process of Example 5 of the present invention.

FIG. 16 is a diagram showing a schematic configuration of a conventional inspection device.

FIG. 17 is a side view of the main parts of the apparatus shown in FIG.

FIG. 18 is an explanatory diagram showing a state in which a tongue sticking out has occurred at the can lid winding tightening portion.

FIG. 19 is an explanatory diagram of a light irradiation state of a normal can lid winding tightening portion.

FIG. 20 is an explanatory diagram of a light irradiation state of the can lid winding tightening portion having a tongue sticking out.

FIG. 21 is an explanatory diagram of an image processing process of a conventional inspection device.

FIG. 22 is an explanatory diagram of an image processing process of a conventional inspection device.

FIG. 23 is an explanatory diagram of an image processing process of a conventional inspection device.

[Explanation of symbols]

 38 Image Processing Device 60 Tongue Out Edge Detection Processing Unit 61 Virtual Clasp Processing Unit 62 Tongue Out Descent Detection Processing Unit 81 Clasp Detection Unit 82 Clasp Pattern Memory Unit 83 Clasp Creation Unit 91 Clasp Detection Unit 92 Multi-Valued Image Processing Unit 93 Tongue Detection Department

Claims (5)

[Claims] 1. An illuminating device for irradiating light from an obliquely upper outer side of a can to be inspected, a photographing device for photographing an image near a can lid winding portion (base) from the side of the can, and the photographing device. From the obtained image, a tongue out edge detection processing unit that detects whether or not there is a bright portion indicating the edge of the tongue out in the dark part of the boundary between the can lid winding part (base) and the can body part, and the can body part Is a colored and bright can, and a virtual can processing unit for creating a virtual cap is provided. 2. An illuminating device for irradiating light from an obliquely upper outside of a can to be inspected, a photographing device for photographing an image near the can lid winding part (base) from the side of the can, and a can body part. In the case of a white can, a defective can inspection device is provided with a tongue-out fall detection processing unit for image-processing a density signal of a multi-valued image from an image obtained by the photographing device. 3. The defective can inspection device according to claim 2, further comprising a virtual base processing unit that creates a virtual base when the can body is a colored and bright can or a white can. 4. An illumination device for irradiating light from obliquely above the outside of a can to be inspected, a photographing device for photographing an image near the can lid winding portion (base) from the side of the can, and the photographing device. From the obtained image, the tongue out edge detection processing unit that detects whether or not there is a bright portion indicating the edge of the tongue out in the dark part of the boundary between the can lid winding part (base) and the can body, and the can body A defective can inspection device comprising: a tongue drip detection processing unit for image-processing a density signal of a multi-valued image from an image obtained by the photographing device when a part of the can is white. 5. A tongue dip detection processing unit for image-processing the density signal of a multi-valued image when the can body is a colored and bright can or a white can.
Defective can inspection device described.