JPH0731137B2 - Inner / outer surface inspection device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the outer appearance of wrinkles, wrinkles,
The present invention relates to a can inner / outer surface inspection device for inspecting an inner surface shape and the like.

[0002]

2. Description of the Related Art In the manufacturing process of cans such as beer and soft drinks, conventionally, in order to inspect the quality of the cans, the inner surface of the cans is inspected for wrinkles and scratches by using a plurality of cameras. I was trying to do it.

Further, as a method for inspecting the quality of a printing state or the like, conventionally, a method of taking a picture of the printing state or the like with a camera and performing a comparative inspection by image processing thereof has been generally adopted. This kind of image inspection method is a method of inspecting an object to be inspected by a camera, processing the image data by an image processing means and inspecting it, and storing an originally correct image as a reference image. The image for inspection (inspection image) is compared with the reference image, and the image of the difference between the two is extracted for inspection.

That is, as shown in FIG. 12, an image (c) of the difference between the inspection image (b) and the reference image (a) is extracted, and the image (c) of this difference is taken out as an inspection output. The defect of the inspection image is inspected.

[0005]

By the way, in the conventional can inner / outer surface inspection device, there is no device for automatically inspecting a printing defect printed on the outer surface of the can, and it relies on visual inspection. Is the current situation. The main reason for this is that the can has a cylindrical shape and inspection cannot be performed in one visual field. Further, in a normal image comparison inspection, in the case of an object having clear characters or patterns such as a printed matter, it is impossible in reality that the reference image and the inspection image completely match even if the inspection image is correct. As shown, as a result of comparing the correct inspection image (b) with the reference image (a), a contour line remains in the image (c) of the difference, so that it is difficult to extract fine defects.

In view of the above-mentioned conventional problems, the can inner / outer surface inspection device of the present invention enables inspection by image processing by taking an image of its appearance with a camera, although it is cylindrical.
To provide a can inner / outer surface inspection device capable of reliably processing an unstable pixel portion of an inspection image to extract only a defective portion of the inspection image, performing the inspection at a high speed, and being configured with a simple circuit. It was made for the purpose.

[0007]

In order to solve the above-mentioned problems, a can inner / outer surface inspection apparatus of the present invention comprises an inspection table provided with a rotating table for rotating the can, and a can for feeding and sending the can to this inspection table. Conveyance means, a visual inspection camera that faces the inspection table and takes an image by dividing the circumferential direction of the can to inspect the external appearance of the can, and controls the rotation of the rotary table for the divided imaging by the camera. Drive means, a mark detection sensor for detecting a reference point mark printed in advance on the can, a control means for emitting a trigger signal by the detection of the reference point mark by this sensor, and imaging the camera and driving the drive means, An image processing comparison means for picking up an image of a non-defective product in advance for each circumferential division of the can by the camera, storing the image as a reference image, comparing it with the inspection image, and inspecting its quality is provided. Than it is.

Further, in the image processing apparatus of the present invention, when the image of a non-defective product is picked up in advance and stored as a reference image, the input image is expanded or contracted and stored, and the inspection image is stored. When comparing with the reference image, the unstable pixel portion of the contour portion of the inspection image is included or excluded to perform comparison.

[0009]

According to the present invention, with the above-described structure, the can is carried into the inspection table, and the can is rotated by the rotating table to detect the reference point mark previously marked on the can. The can is rotated by the turntable starting from, and the appearance of the cylindrical can is divided by the appearance inspection camera in the circumferential direction by the rotation control of the turntable by the control means, and the images are taken at each divided position. By comparing the inspection image of the can appearance with the reference image, the quality of the can can be inspected.

Further, when the reference image is stored by the image processing comparison means, the input image is expanded or contracted and stored, and when the inspection image is compared with the reference image, the contour portion of the inspection image is not detected. By including the stable pixel portion or excluding the stable pixel portion for comparison, it is possible to reliably process the unstable pixel portion of the inspection image and extract only the defective portion of the inspection image.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A can inner / outer surface inspection device according to an embodiment of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram showing its outline.

In the figure, 1 is a can, and a can carrying-in section 2
The can 1 carried in is rotated by a turntable 4 for rotating the can 1.
Are supplied to the inspection table 3 provided with. The can 1 is inspected on the inspection table 3 as will be described later.
It is sent out from the can carrying-out section 6. On the inspection table 3 portion, a visual inspection camera 7 that faces the loaded can 1 and divides the circumferential direction of the can 1 to take an image to inspect the external appearance of the can 1 is printed in advance on the can 1. Mark detection sensor 8 for detecting a reference point mark, an inner surface inspection camera 9 provided for inspecting the inner surface of the can 1, and a glossiness for imaging the entire circumference of the can 1 for inspecting wrinkles of the can 1. And a wrinkle detection sensor 10 composed of a meter.

The rotating table 4 provided on the inspection table 3 is adapted to rotate at a high speed for the divided image pickup by the appearance inspection camera 7, and the motor 11 controls the rotation. , The driving force of this motor 11 is the belt 13 connected to each rotary table 4 from the belt pulley 12.
Is transmitted through. Here, the motor 11 drives the rotary table 4 to rotate to rotate the can 1 itself, but the inspection table 3 itself rotates appropriately and intermittently by a separately provided drive means such as a motor. Is.

The appearance inspection camera 7 is provided with, for example, five units (7a to 7e), and is used to inspect the can 1 by dividing the field of view into five equal parts in the circumferential direction. Each of the cameras 7a to 7e is shown in FIG. As shown in the drawing, the appearance of the can 1 divided into five equal parts (72 °) is imaged, and the image signals thereof are supplied to the image processing comparison means 15 having the corresponding image processing comparison means 15a to 15e. It has become so.

Here, the image processing comparison means 15a-15
An image of a non-defective product is captured in advance for each e, and this is stored as a reference image, and the quality of the image is compared with the inspection image for inspection. The image processing comparison means 15a to 15e will be described in detail later. In addition, when the image of a non-defective product is captured in advance and the input image is subjected to expansion or contraction processing and stored as a reference image, the image is stored, and the contour portion of the inspection image is compared when the inspection image is compared with the reference image. The unstable pixel portion is included or excluded for comparison. In addition,
The image by the image processing comparison means 15 can be confirmed by the monitor 16.

Reference numeral 17 is a reflection box, and the camera 7
The image is illuminated on the can 1 when the can 1 is imaged by the. The illumination by the light source 18 by a plurality of DC-lit incandescent bulbs is indirectly performed by the illumination of the reflection box 17 as shown in FIG. The reflective box 17 uses a white plate. This is because, in the appearance inspection of the can 1, the cylindrical can 1 which is difficult to be uniformly illuminated for obtaining a stable image is not directly illuminated, but is uniformly illuminated in the field of view of the camera. This is for displaying this on the surface of the can 1, and the print on the surface of the can 1 is captured as a clear image.

Further, the reflection box 17 includes the side plate 1
7a is provided so that the reflection box 17 is projected in the visual field range of the camera 7 as shown in FIG.
The reflection box 17 may be a transmission illumination plate that can diffusely reflect, for example, milk white, instead of the reflection illumination plate generated by the light source 18.

The mark detection sensor 8 is shown in FIG.
Reference point mark 1a printed on the can 1 as shown in Fig.
The detected sensor output is sent from the controller 8a to the control means 14 including the CPU.
Upon detection of the reference point mark 1a by the mark detection sensor 8, the control means 14 issues a trigger signal to start image pickup of the cameras 7 and 9, start detection by the wrinkle detection sensor 10 and drive control of the motor 11. It is a thing. Reference point mark 1a by this mark detection sensor 8
4B, the corresponding portion of the mark 1a is spot-illuminated by the fiber 8b connected to the halogen light source, and the sensor 8 detects it.

The inner surface inspection camera 9 provided for inspecting the inner surface of the can 1 is, as shown in FIG. 5, viewed from obliquely above the inner surface of the can 1 with a one-dimensional sensor from the mouth of the can 1. One line up to the bottom is sequentially processed, and the processing of the imaged signal is performed by the image processing comparison means 19. The processing is performed by rotating the can 1 to obtain data from the previous line for each line. The comparison is made, and if a change occurs, it is judged as a defective product.

Here, since the inner surface of the can 1 is basically in the same state over the entire circumference, the defective product can be identified only by detecting the change. If a line sensor is used as the camera 9 for inspecting the inner surface, only one line needs to be illuminated, so that the illumination can be simplified and the period of one line scan of the line sensor can be adjusted to the horizontal synchronizing signal of the two-dimensional camera. For example, it is possible to display the image of the line sensor on the TV monitor. In addition, for the illumination of the imaging portion of the can 1, it is possible to illuminate a necessary portion on the inner surface of the can 1 by the fiber ring 9a provided by the halogen light source provided in front of the camera 9.

The wrinkle detecting sensor 10 is composed of a gloss meter, and is provided to inspect wrinkles on the entire circumference of the neck portion of the can 1 as shown in FIG. It is processed by the image processing means 20.

Next, an example of a detailed structure of the inspection table 3 portion will be described with reference to FIGS. The inspection table 3 is provided with a plurality of turntables 4 on the outer circumference thereof, the turntable 4 is rotatable on the first mounting plate portion 3a of the test table 3, and the grooves 4a are formed on the outer circumference of the turntable 4. The belt 13 is fitted into the groove 4a, and the rotary table 4 is rotated by driving the belt 13 by the motor 11. Further, in order to enable the cans 1 laterally supplied to the inspection table 3 to be mounted on the turntable 4, the turntable 4 has a through hole 4b at its center.
Is provided, and the can 1 is sucked by a vacuum (not shown) through the through hole 4b. Therefore, the can 1 rotates with the rotary table 4 while being attracted to the rotary table 4.

On the inspection table 3, guide rollers 21a, 21b and 21c are further provided with second and third mounting plate portions 3b and 3c in order to rotate the can 1 more stably when the can 1 is rotated by the turntable 4. It is rotatably installed. It should be noted that the first to third mounting plate portions 3 a to 3 c are attached to the shaft 22 with the key 2
The shaft 2 is mounted with its rotation direction restricted by
The rotation of the inspection table 3 is performed by the rotation of 2.

As shown in FIGS. 7 and 9, the turntable 5 is provided with a plurality of can receiving recesses 5a around it, and the cans 1 are vacuumed (not shown) in the can receiving recesses 5a. An adsorber 24 for adsorbing is provided. As described above, by providing the suction device 24, when the can 1 is carried out to the can carry-out section 6 on the turntable 5, it can be easily carried out to the can carry-out section 6 even though it is once directed downward. Can be done. Moreover, at that time, the adsorber 24
The can 1 can be easily dropped and carried out by canceling the suction by.

Next, a detailed example of image processing by the individual image processing comparison means 15a to 15e in the image processing comparison means 15 will be described. In this image processing comparison means, when a non-defective image is captured in advance and stored as a reference image, the input image is subjected to expansion or contraction processing and stored, and when the inspection image is compared with the reference image, the inspection is performed. The comparison is performed by including or excluding unstable pixel portions in the contour portion of the image. Hereinafter, a specific example of the image processing will be described with reference to FIGS.

FIG. 10 shows the relationship between the input image of the reference image and the stored reference image and inspection image according to the present invention, where A is the input image of a predetermined image which should be the object to be inspected. The image obtained by expanding the unstable pixel portion of the contour portion of the input image A is the first reference image A1, and conversely, the image obtained by performing the contraction processing of the unstable pixel portion of the contour portion of the input image A is the second reference image A2. And the inspection image B
1 and B2.

Here, the first reference image A1 is a black expanded image (an image whose brightness is lower is expanded), and this is effectively detected when the inspection image B1 has a black defect C1. It is a thing. On the other hand, the second reference image A2 is a white expanded image (an image with a higher brightness is expanded),
When the inspection image B2 has a white defect C2, this is effectively detected. In the description of FIG. 10, the image is described by the Roman character “A” for convenience, but it is not related to the actual image.

FIG. 11 is a block diagram showing an embodiment for carrying out the image inspection processing shown in FIG. 10, and this embodiment will be described below. In FIG. 11, the image captured by the camera 7 is sent as an input image to the frame memory 31 of the image processing comparison means 15. The input image is digitized and stored in the frame memory 31, and the memory can be freely read and written by the CPU 32.

Reference numeral 33 is a black defect detecting portion, which is the first portion described above.
A first subtraction board 35 that stores the reference image A1 of FIG. 1 and outputs the difference from the inspection image B1, and a binarization counter 36 that binarizes the output of the subtraction board 35 and counts the number of pixels that have exceeded the set level. Is equipped with. The white defect detection unit 34 stores the second reference image A2 described above and outputs the difference from the inspection image B2.
7 and a binarization counter 38 that binarizes the output of the subtraction board 37 and counts the number of pixels whose set level is exceeded. The binarization counter 36 or 3
By selecting the output of No. 8 by the changeover switch 39 and outputting to the monitor 16, the difference between the inspection image and the reference image can be confirmed.

Reference numeral 40 is a synchronizing signal generating means, which generates a synchronizing signal necessary for image processing and supplies it to each board or the like via the CPU 32. Reference numeral 41 is an input / output means for inputting operation signals and outputting control signals to the outside. By such image processing,
Even if the inspection object is a normal product and has an unstable pixel portion in the contour portion, this unstable pixel portion can be included or excluded by the reference image, and only the portion that should truly be a defect can be detected. .

Next, the overall operation will be described. First, the can 1 conveyed from the can carry-in section 2 is carried into the inspection table 3 and is moved to a position facing the can appearance inspection camera 7 by the rotation of the inspection table 3, where the can 1 is subjected to a predetermined inspection. Each will be done.

That is, the can 1 is rotated by the turntable 4 which is rotated by the drive of the motor 11 via the belt 13, and the mark detection sensor 8 detects the reference point mark 1a of the can 1 to cause the control means 14 to operate. Sends a trigger signal. When the control unit 14 captures the first image from the first visual inspection camera 7a in response to the trigger signal, it outputs "7" to the encoder pulse counter as a control signal for the motor 11.
Set so that an interrupt signal is output at 2 °. If an interrupt occurs, the second image from the second visual inspection camera 7b is captured, and the encoder pulse counter again displays "72".
Set so that an interrupt signal is output at ゜ ”
It will be repeated twice.

In this way, the outer appearance of the can 1 is divided into five parts while the rotating base 4 is rotated by the motor 11, and the can 7 is divided into five parts.
.About.7e, images are sequentially picked up at high speed, and the captured images are compared and inspected by the image processing comparing means 15 described above. Then, if there is a defective product as a result of the comparison inspection, the information is stored in the CPU of the control means 14.

As described above, the can 1 is subjected to the inner surface inspection of the can 1 by the inner surface inspection camera 9 and the wrinkle detection by the wrinkle detection sensor 10 during the rotation by the rotating table 4 as described above. If there is a defect as a result of the processing by the processing means 19 and 20, the information is stored in the CPU of the control means 14.

The can 1 which has been inspected by the inspection table 3 is adsorbed to the turntable 5 and supplied to the can unloading section 6 side. Here, the position of the turntable 4 for detecting defective products of the can 1 on the inspection table 3 and the suction position of the defective products on the suction device 24 of the turntable 5 are stored in the CPU of the control means 14. By
It can be identified at the time of carrying out from the turntable 5 to the can carrying-out section 6. For example, the can carry-out section 6 is divided into upper and lower carry-out sections 6a and 6b, and the timing of release of suction by the suction device 24 is shifted to carry out good products to the carry-out section 6a and defective products to the carry-out section 6b, respectively. You can

[0036]

As described above, according to the present invention, the inspection table provided with the rotary table for rotating the can, the can conveying means for supplying and delivering the can to the inspection table, and the can table facing the inspection table are provided. A camera for visual inspection that divides the circumferential direction of the can to take an image for visual inspection, a driving unit that controls the rotation of the rotary table for the split imaging by the camera, and a reference point printed on the can in advance. A mark detection sensor for detecting a mark, a control unit for emitting a trigger signal by detecting a reference point mark by the sensor to drive the camera to capture an image and drive a driving unit, and a non-defective product in advance for each circumferential division of the can by the camera. It is equipped with an image processing comparison means for picking up an image of the image, storing it as a reference image, comparing it with the inspection image, and inspecting whether the image is good or bad. Not its appearance can be made possible inspection by to image processing captured by the camera, and in which a test can be performed at high speed.

Further, in the image processing apparatus of the present invention, when the image of a non-defective product is picked up in advance and is stored as the reference image, the input image is expanded or contracted and stored, and the inspection image is stored. Since the unstable pixel portion of the contour portion of the inspection image is included or excluded when the comparison with the reference image is performed, the unstable pixel portion of the inspection image is surely processed. It is possible to extract only the defective portion of the inspection image, and it is possible to configure with a simple circuit.

[Brief description of drawings]

FIG. 1 is an overall configuration diagram showing an embodiment of the present invention.

FIG. 2 is an explanatory diagram of an example of split imaging of a can by the appearance inspection camera.

3A is a side view for explaining the operation of the reflection box, and FIG. 3B is a plan view thereof.

FIG. 4 (a) is a front view of a can with the same reference point mark,
(B) is a detection state explanatory diagram of the reference point mark.

FIG. 5 is an explanatory view of an inspection state by the camera for inspecting the inner surface of the can.

FIG. 6 is an explanatory view of an inspection state by the wrinkle inspection sensor.

FIG. 7 is a front view showing a specific example of the inspection table and the turntable.

FIG. 8 is a cross-sectional view of the inspection table.

FIG. 9 is a cross-sectional view of the turntable.

FIG. 10 is an explanatory diagram showing a relationship between an input image of a reference image and the stored reference image and inspection image in the image processing comparing unit.

FIG. 11 is a block diagram of the image processing comparison means.

FIG. 12 is an explanatory diagram of a comparative image in a conventional image inspection method.

FIG. 13 is an explanatory diagram of the same problem.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 can 2 can carry-in section 3 inspection table 4 turntable 5 turntable 6 can carry-out section 7 appearance inspection camera 8 mark detection sensor 9 inner surface inspection camera 10 wrinkle detection sensor 11 drive motor 14 control means 15 image processing comparison means A Input image A1 Black expansion image A2 White expansion image B1 Inspection image B2 Inspection image

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-5-126762 (JP, A) JP-A-1-206242 (JP, A) Actual development Sho-61-195458 (JP, U)

Claims (4)

[Claims] 1. An inspection table provided with a turntable for rotating the can, a can conveying means for supplying the can to the inspection table and sending it out, and a peripheral of the can for facing the inspection table and inspecting the appearance of the can. A visual inspection camera that divides and captures images in different directions, a drive unit that controls the rotation of the rotary table for the split image capturing by the camera, and a mark detection sensor that detects a reference point mark printed in advance on a can. A control unit that emits a trigger signal by detecting the reference point mark by this sensor to image the camera and drive the driving unit, and an image of a non-defective product is taken in advance by the camera for each circumferential division of the can and is used as a reference image. A device for inspecting the inside and outside of a can, characterized by comprising image processing comparing means which stores the image as an image and compares it with an inspection image to inspect its quality. 2. The image processing comparison means, when a non-defective image is picked up in advance and stored as a reference image, the input image is subjected to expansion or contraction processing and stored, and the inspection image is compared with the reference image. 2. The can inner / outer surface inspection device according to claim 1, wherein the instability pixel portion of the contour portion of the inspection image is sometimes included or excluded for comparison. 3. The can inner / outer surface inspection apparatus according to claim 1, further comprising a wrinkle detecting sensor including a gloss meter for inspecting the entire circumference of the can to inspect the wrinkle of the can. 4. The can inner / outer surface inspection apparatus according to claim 1, further comprising an inner surface inspection camera provided to inspect the inner surface of the can.