JP3224667B2 - Metal can end winding appearance inspection method and apparatus and image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for inspecting the appearance of a metal can for food, for example, for detecting defective tightening between a can body and a can bottom and between a can body and a can lid.

[0002]

2. Description of the Related Art As a conventional method of inspecting a metal can-fastened portion, an end portion of a metal can conveyed on a conveyor is illuminated with illumination light from above, and a part of the fastened portion protrudes outside the can body. The reflected image of the portion is captured by a television camera arranged in the outer peripheral direction of the end of the metal can, and the captured signal is input to an image processing device and analyzed to determine the quality of the tightened portion. The above known examples can be found in JP-A-4-121647 and JP-A-4-121648.

[0003]

If the tightening is not performed properly at the tightening portion at the end of the metal can, a part of the can body protrudes from the outside, that is, a so-called tongue defect occurs. In the above prior art, inspection is performed by irradiating illumination light from above a metal can to be inspected, and arranging a plurality of television cameras beside the tightened portion. If there is a tongue sticking out part due to insufficient tightening, the image is input to any one of the plurality of television cameras, and the image processing device determines the quality.

FIG. 1 shows an example of the arrangement of the TV camera and the metal can.
It is shown in FIG. In FIG. 11, the metal can comprises a can body 11 and a can bottom or can lid 12, and the can body 11 has a welded seam portion 15 for forming the can body. It is fastened to the can bottom or can lid 12 by a fastening portion 13 (this is called double fastening). One of the examples of poor tightening of the winding part 13 is the presence of the tongue sticking part 14. In the winding part 13, the can bottom or can lid 1
2 and the end of the can body 11 are to be bent so as to be wrapped in the inside, and then wound.
There is a case where the end of 1 is exposed outside without wrapping around inside. The phenomenon in which the end of the can body is exposed outward from the tightened portion is called tongue sticking. Fig. 1
1 is indicated by the symbol 14. This is a phenomenon that appears in a part of the circumferential direction of the winding portion 13 and is usually one example,
There are also two or more examples. When the tongue sticking out portion 14 is generated, the airtightness inside the can is impaired or the appearance is deteriorated, so that it is determined that the tightening is poor.

[0005] In Fig. 11, a TV camera 40 is provided on the side for inspecting the presence or absence of the tongue protruding portion 14. This imaging field of view 4
Reference numeral 1 denotes a range including the wind-up portion 13, and the presence or absence of the tongue out portion 14 is determined by image processing.

[0006] Generally, a metal pattern for a beverage or the like is usually printed with a colorful pattern on the outer periphery to enhance the image of the product. However, as long as the pattern does not reach the vicinity of the can end winding portion 13 and the vicinity of the winding portion is a color close to plain, the above-described conventional technique is effective without any hindrance.

However, as shown in FIG. 11, the pattern 11b on the outer peripheral portion of the metal can is printed so as to reach the tightened portion, and the outer periphery of the metal can is a dark portion 11a, the pattern 11b is a bright portion and the contrast is high. However, there is a problem that a reflected image from a pattern is input and cannot be distinguished from an image of a tongue sticking-out part.

In the above-mentioned prior art, a plurality of television cameras are provided in order to input the image of the end portion of the metal can to the TV camera from the side. Or rotating the metal can in front of one TV camera, and inputting the data in a plurality of times.

SUMMARY OF THE INVENTION The object of the present invention is to solve the above-mentioned problems, and it is possible to use a metal can having a plain body, a pattern-printed can, and its color. Regardless, an object of the present invention is to provide a metal can end winding appearance inspection method and apparatus and an image processing method for easily and inexpensively determining the quality of a metal can end winding portion.

[0010]

According to the present invention, in a metal can end winding appearance inspection method, a reflection image of an upper end of a winding by an illumination light from a ring type lighting device arranged above a winding portion is formed. An image is taken by a television camera arranged coaxially with the center of the lighting device, the input image is converted into a digital multi-tone image, a ring-shaped image at the upper end of the tightening is obtained, and a ring width is radially formed from the center of the ring-shaped image. Is measured at appropriate intervals over the entire circumference, and when each of the ring width dimensions is larger than a preset threshold value, it is determined that the product is defective.

Further, in the present invention, in a metal can end winding appearance inspection method, a reflection image of an upper end of the winding by an illumination light from a ring type lighting device arranged above a winding portion is defined as a center of the lighting device. An image is captured by a television camera arranged coaxially, the input image is converted into a digital multi-tone image, a ring-shaped image of the upper end of the tightening is obtained, and the ring width is radially spaced from the center of the ring-shaped image at appropriate intervals. The entire circumference is measured, and the upper and lower thresholds are set for the ring width dimension. If the ring width dimension exceeds the upper threshold, it is determined to be defective. Disclosed is a metal can end winding appearance inspection method characterized in that when two or more width dimensions are present and the position is located at a position apart from the center angle α degrees of the ring-shaped image, it is determined to be defective.

Further, in the present invention, the threshold value for the ring width dimension is an arithmetic constant obtained by multiplying the average value of the ring width dimension by an appropriate value.

Further, according to the present invention, the absolute value of the dimensional difference between adjacent measurement points of the ring width dimension is determined, and the pass / fail judgment is made.

Further, in the present invention, the threshold value for the absolute value of the dimensional difference between adjacent measurement points of the ring width dimension is an arithmetic constant obtained by multiplying the average value of the absolute values of the dimensional differences by an appropriate value.

Further, according to the present invention, the metal can is conveyed on a conveyor, and the sensor detects that the metal can enters the imaging range of the TV camera, and an image is taken by the TV camera based on the detection signal. I made it.

Further, in the present invention, in a metal can-end winding appearance inspection apparatus, a ring-type illumination device disposed above a winding portion and an image of a reflection of an upper end of the winding by the illumination light are taken.
A television camera disposed coaxially with the center of the lighting device, and means for converting the captured image into a digital multi-tone image;
A metal can end winding appearance inspection apparatus comprising: a ring-shaped image of the upper end of the winding in this image, which is sampled in the circumferential direction, and monitoring the magnitude of each ring width to determine whether the winding is good or defective. Is disclosed.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described with reference to FIGS. FIG. 1 shows an apparatus configuration of the present invention. A ring-type illumination device 2 is disposed above a metal can 10 conveyed on a conveyor 1, and a television camera 4 is disposed coaxially with the illumination device 2. Set up. The ring-type lighting device 2 includes a
Is illuminated from above, and the entire ring 13 is illuminated from the entire ring. The TV camera 4 captures an image of the fastening portion 13 from a space inside the ring. By arranging the TV camera 4 and the lighting device 2 coaxially, the TV camera 4 uniformly captures an image of the entire tightening unit 13 from above.

The metal can 10 conveyed on the conveyor 1 in the direction of the arrow is detected by the sensor 5 at a moment when the metal can 10 is directly below the lighting device 2 and the television camera 4, and this detection signal is inputted to the image processing device 6, The illumination device 2 emits light via the illumination power source 20 and irradiates the tightening unit 13. At the same time, the detection signal of the sensor 5 activates the television camera 4 via the image processing device 6 to cause the image of the contour image of the tightening portion 11 to be captured, and then input the image to the image processing device 6. The result of the discrimination processing by the image processing device 6 is sent to the sorting device 7, where the non-defective product and the defective product are sorted.

Referring to FIG. 2, the internal structure and basic operation of the image processing apparatus will be described. The input signal of the television camera 4 is input to the image processing device 6, converted into a digital multi-tone image by the A / D converter 601, and stored in the image memory 602. The image measurement calculation processing unit 603 measures and calculates the size of the ring-shaped image of the tightened portion from the digital multi-tone image. The threshold value setting unit 604 sets a limit value for quality determination with respect to the size of the ring-shaped image.
05 memorize. The comparison operation processing unit 606 compares the value measured by the image measurement operation processing unit 603 with the value of the threshold value memory 605 to determine the quality, and outputs a pass / fail signal to an external sorting device or the like.

Next, typical improper tightening which occurs at the end of the metal can at the end of the metal can will be described with reference to FIGS. FIG.
FIG. 4 is a perspective view showing an upper end portion of the metal can, and a can body 11 of the metal can.
Is wound and sealed like a can bottom or can lid 12 and a wrapping portion 13. Poorly tightened portions generally appear as a portion of the can body 11 below the tightened portion 13, like a tongue protrusion 14.

The can body 11 may or may not have a weld seam 15 depending on the manufacturing method. Weld seam 15
Some are called three-piece cans and others are called two-piece cans. In this embodiment, the welding seam portion 15 exists 3
Piece cans are to be inspected.

FIG. 4 shows a cross section of the tightened portion shown in FIG. 3. FIG. 4 (a) is a cross section taken along the line AA of FIG. 3, and similarly, FIG. FIG. 4C is a diagram showing a CC section. FIG. 4A shows a normal tightening state, in which the can body 11 and the can bottom or the can lid 12 are tightly wound as shown in FIG. Agent 1
6 work in becomes the sealed state, the width d ₁ of the seamed upper end is captured by a TV camera as a high luminance image as reflected light.

FIG. 4B is a cross-sectional view of the vicinity of both side ends of the tongue sticking out portion 14 of FIG. 3. The can body 11 and the can bottom or can lid 12 are always bent incorrectly, and the brightness of the upper end portion of the winding-up portion is reduced. The width d ₂ of the high reflection part is expanded as shown in the figure. This abnormal bending is various in various ways, and is not limited to the illustrated embodiment. FIG. 4C is a cross-sectional view of the vicinity of the center of the tongue extending portion 14 in FIG. 3 and shows an example in which a part of the can body 11 is not normally caught and protrudes outside. The width d ₃ is substantially equal to d ₁ in FIG.

Here, the appearance of FIGS. 4A, 4B and 4C will be described. When the tongue sticks out, the start part and the end part on the circumference of the tongue stick part become wider when viewed from above, and there is usually no significant spread at the intermediate part. The start part and the end part are examples shown in FIG. The intermediate portion is the example shown in FIG.

FIG. 5 shows an image obtained by A / D-converting a video image of the tightened portion shown in FIG. 3 by a television camera from above with the method shown in FIG. In the example where the image is converted into a gradation image, if only the reflected light with high luminance from the upper end of the tightened portion is expressed, a ring-shaped image is obtained as shown in FIG.

In FIG. 5, the ring-shaped image portion 111 is a reflection image of the upper end viewed from the TV camera 4 of the winding portion 13;
This almost indicates the shape of the tightening portion 13. Video unit 14
a, 14b in FIG. 3, FIG. 4 lingual out aspects of B-B of (b) (beginning and end portion) of the image (14a, the width of 14b d _2), the video portion 14c in FIG. 3, FIG. 4 ( c) The image of the tongue sticking out middle part in CC (the width of 14c is d ₃ ), and the image part 14d is the image of the normal part in AA in FIG. 3 and FIG. 4A (the width of 14d is d _1). ). 15a is an image of the seal portion 15. This image is caused by an increase in the overall thickness due to the overlap of the can body plates.

As described above, in the tongue protruding portion 14, images having a width of d ₂ appear on both sides. On the other hand, a similar image 15a also appears on the seal portion 15. Therefore, the discrimination of the tongue sticking out part 14 is performed by 14a and 14
b must be detected and distinguished from 15a.

The images 14a and 14b on the two side surfaces of the tongue protruding portion 14 appear empirically at a predetermined central angle α or more. The central angle α is the angle of each central position of the images 14a and 14b. The size of the central angle α is determined by the diameter of the can, and in the case of soft drink cans, α is about 20 °. In an example where α <20 °, images such as 14 a and 14 b in FIG.
Does not occur. Therefore, in such a case, it is not considered that the can is poorly wound.

In FIG. 5, the angle θ from the reference line is increased at regular intervals, and the width dimensions of the ring-shaped image at that time are sequentially plotted and graphed as shown in FIGS. Shows the angle θ, and the vertical axis shows the ring width dimension d _i or d ′ _i .

FIG. 6 is a plot value in the case as shown in FIG. 5, and shows an example in which three peaks P ₁ , P ₂ and P ₃ appear. The peaks P ₁ and P ₂ are generated due to the thickness d ₂ of both side portions of the tongue projection 14, and the peak P ₃ is the weld seam 1
This is caused by a thickness d _{4 of} 5. Here, the peaks P ₁ and P ₂ occur at an interval equal to or greater than the center angle α, and the peak value due to the welding seam portion 15 is the peak P ₁ ,
The feature is that the value is smaller than P ₂ .

[0031] Therefore, detects the peak P ₁ and P _2, it is necessary not to determine the peak P ₃ with similar peak and peak P _1, P _2. Therefore, the peak P ₁
It sets a larger level threshold value J ₁ than P ₂ and, P
₁ , a level threshold J ₂ smaller than P ₂ but larger than P ₃
Set. Further, the center angle α is set as a peak interval angle threshold. By monitoring the peaks that appear with these parameters J ₁ , J ₂ and α, the peaks P ₁ and P ₂ can be detected,
And P ₃ is enabled to exclude from detection target.

FIG. 7 shows another plot example of the image as shown in FIG. FIG. 6 shows an example in which the image values detected for each sample angle are plotted as they are, but FIG. 7 is an example effective when uniform brightness cannot be obtained as a whole due to the conditions of the illumination light and the like. taking the absolute value of the difference between the video values of the sample angle each fit, it is obtained taking the absolute value d _i of the difference on the vertical axis. According to FIG. 7, the peaks P ₁ , P ₂ ,
P ₃ becomes P ₁ ′, P ₂ ′, P ₃ ′, and the level threshold J ₁ ′,
The peaks P ₁ ′ and P ₂ ′ can be detected by setting J ₂ ′ and the central angle α as threshold parameters.

FIG. 8 is an example of a two-piece can in which image values at each sample angle are plotted. Similarly, when a tongue sticks out, two P ₁ and P ₂ appear. However, since a two-piece can, the peak P ₃ is absent by weld seam. Therefore, P ₁ and P ₂ can be detected by setting a single level threshold value J ₀ and setting the center angle α as an angle threshold value.

FIGS. 9 and 10 show detailed diagrams of the determination processing flow in FIGS. 6 and 7. FIG. In FIG. 9, the threshold values J ₁ and J ₂ are values calculated by the following equations.

J ₁ = 3 × d _im J ₂ = 2 × d _im The threshold values J ₁ ′ and J ₂ ′ in FIG. 10 are values calculated by the following equations.

## EQU2 ## J ₁ ′ = 4 × d ′ _im J ₂ ′ = 3 × d ′ _im Here, _dim is a plot of FIG. 6 and d ′ _im is the average value of all sample values for 360 °. 360 in the plot of FIG.
It is the average value of all sample values of ゜. Naturally, there may be other numerical examples other than the average value.

FIG. 9 shows an example of the processing flow of FIG. 6, in which the signal input in step 610 is
Is converted to a digital multi-tone image by A / D conversion. Since the digital multi-tone image is a ring-shaped image as shown in FIG. 5, the center point of the ring image is determined in step 612, and the ring width d _i is determined at step 613 at an arbitrary sample angle from the center point. Are sequentially measured, and step 61
5 of the determination unit, compared with a threshold value J ₁ which is set in advance in step 614, and d _i ≧ J ₁ if defective, d _i <
If J ₁ , the judgment unit of step 616 compares it with J ₂ set in advance in step 614, and if d _i <J _2, it is determined to be good. If d _i ≧ J ₂ , step 609 is performed.
Is sent to the determination unit. In the determination unit of step 609, d
_It is determined whether or not there is a portion satisfying d _i ≧ J _{2 at} a position of θ where _i ≧ J ₂ is detected and at a position separated by α degrees or more from the position of θ. Good product.

FIG. 10 is an example of the processing flow of FIG. 7, and the processing flow of steps 610 to 613 is the same as that of FIG.
Is the same as In the calculation section of step 620, d ′ _i is obtained from d _i obtained above by the following equation.

## EQU3 ## d ′ _i = | d _i −d _i−1 | In the determination section of step 622, step 621 is performed in advance.
Is compared with the threshold value J ′ ₁ set as described above. If d ′ _i ≧ J ′ ₁ , the product is determined to be defective. If d ′ _i <J ′ ₁ , the process proceeds to the determination section of step 623, and is set in advance to step 621. 'compared to _2, d' is J 'is a ₂ if good, if, d' _i <J when the _i ≧ J _'2 are sent to the determination of step 624. In the determination section of step 624, d ′ _i ≧ J ′ ₂
It is determined whether or not there is a portion satisfying d ′ _i ≧ J ′ _{2 at} the position of θ where the angle θ is detected and at a position separated by α degrees or more from the position of θ. I do.

9 and 10, the threshold values J ₁ , J ₂ ,
J ′ ₁ and J ′ ₂ are operation type constants as shown in Expressions 1 and 2, but by using the operation type constants as thresholds in this way,
Although the tolerance range for fluctuations in the brightness of the input video increases,
Regardless of the present example, an appropriate numerical value corresponding to the above-described threshold value may be determined by a sample test, and the numerical value may be directly input.

In this embodiment, an example of processing on a digital multi-tone image has been described. However, if a stable image can be obtained, an input image may be binarized to perform the same discrimination processing as described above. Included in the invention.

There are other processing methods. J ₁ larger than J ₂
Smaller than the are examples according to the tongue out unit, without taking into account the central angle alpha, sometimes it can be determined whether the tongue out in this J ₁ and J ₂ is generated. Further, when the tongue out portion is present, will be present two larger peaks than always J _1, greater than J ₁ J ₂
Monitoring of tongue sticking out is possible depending on whether two smaller peaks exist. In particular, when a tongue protruding portion occurs at two or more locations, the number of peaks is always an even number of four or more, and it can be determined based on whether or not this peak appears. Also, on the three-piece side where the welded seam exists, there is a difference in the peak value, but since there is one peak due to the welded seam, the peak is always an odd number.

There is also an exceptional case in which the entire tongue-out portion has a predetermined width at the tongue-out portion (d _{3 in} FIG. 5 is the same as d ₂ ). To detect this tongue sticking out, if the difference shown in FIG. 7 is taken, a steep peak appears only on the side surface of the tongue sticking out part, which can be monitored by the flow of FIG.

Although an image pickup means other than a TV camera may be used, and a ring-type illumination device is used, a ring-shaped image can be realized with a normal illumination device by rotating the can itself.

[0042]

According to the present invention, a defective fastening which occurs in a tightened portion of a metal can is imaged by a television camera disposed above the tightened portion, and the quality is determined by image processing. Therefore, the influence of the reflection from the pattern portion printed on the can body can be avoided, and the quality of the metal can end winding portion can be determined reliably. In particular, in a food can or the like, it is effective as a means for preventing in advance the adverse effect that improperly sealed cans leak to the market due to improper winding. Also,
Compared to the method of taking an image with a plurality of television cameras from the outer peripheral direction of the metal can, there is an advantage that the apparatus is simple and the installation space is small.

[Brief description of the drawings]

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

FIG. 2 is a configuration diagram of an image processing apparatus of the present invention.

FIG. 3 is a perspective view of a metal can end winding portion.

FIG. 4 is an explanatory view showing a cross section of a wound portion.

FIG. 5 is a diagram showing a digital multi-tone ring-shaped image obtained by capturing an A / D-converted image of a metal can end winding portion from above.

FIG. 6 is a graph in which ring width dimensions of a ring-shaped image of a tightened portion are plotted.

FIG. 7 is a graph in which a ring width dimension of a ring-shaped image of a tightened portion is plotted.

FIG. 8 is a graph in which a ring width dimension of a ring-shaped image of a tightened portion is plotted.

FIG. 9 is a diagram illustrating a determination processing flow of image processing.

FIG. 10 is a diagram illustrating a determination processing flow of image processing.

FIG. 11 is a diagram illustrating a problem of a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Metal can 2 Illumination device 3 Conveyor 4 TV camera 5 Sensor 6 Image processing device 7 Sorting device 11 Can body 12 Can lid 13 Winding part 14 Tongue out part 15 Weld seam 16 Sealant 601 A / D conversion part 602 Image memory 603 Image measurement calculation processing section 604 Threshold setting section 605 Threshold memory 606 Comparison calculation section 610 Video input step 611 A / D conversion step 612 Center measurement step 613 Ring width measurement step 614 Threshold 615 Comparison judgment step 616 Comparison judgment step 617 Comparison judgment step 620 Ring width difference calculation step 621 Threshold value 622 Comparison judgment step 623 Comparison judgment step 624 Comparison judgment step

Continuation of the front page (56) References JP-A-5-93699 (JP, A) JP-A-4-121647 (JP, A) JP-A-5-145833 (JP, A) (58) Fields investigated (Int) .Cl. ⁷ , DB name) G01N 21/90-21/952 G01B 11/08 G01B 11/24 G06T 1/00 400 G06T 7/00 300

Claims (8)

(57) [Claims] In a method for inspecting the appearance of a metal can end winding,
A reflected image of the upper end of the winding by the illumination light from the ring-type illumination device arranged above the winding unit is captured by a television camera arranged coaxially with the center of the illumination device, and the input image is a digital multi-tone image. To obtain a ring-shaped image of the upper end of the winding, radially measured the ring width at appropriate intervals radially from the center of the ring-shaped image, when each ring width dimension is larger than a preset threshold, A method for inspecting the end of a metal can end by tightening, which is determined to be defective. 2. A method for inspecting the appearance of a metal can end tightening,
A reflected image of the upper end of the winding by the illumination light from the ring-type illumination device arranged above the winding unit is captured by a television camera arranged coaxially with the center of the illumination device, and the input image is a digital multi-tone image. To obtain a ring-shaped image of the upper end of the winding, radially measure the ring width at appropriate intervals radially from the center of the ring-shaped image, for the ring width dimension, set the upper and lower thresholds If there is the above-mentioned ring width dimension exceeding the upper limit threshold, it is determined to be defective, there are two or more ring width dimensions exceeding the lower limit threshold and less than the upper threshold, and the position is the central angle α of the ring-shaped image. A metal can end tightening appearance inspection method characterized by determining that the metal can is defective when the metal can is located at a distance apart. 3. The external appearance of the metal can end tightening according to claim 1, wherein the threshold value for the ring width dimension is an arithmetic constant obtained by multiplying an average value of the ring width dimension by an appropriate value. Inspection methods. 4. The method according to claim 1, wherein an absolute value of a dimensional difference between adjacent measurement points of the ring width dimension is determined, and the quality is determined. 5. The arithmetic type constant according to claim 4, wherein the threshold value for the absolute value of the dimensional difference between adjacent measurement points of the ring width dimension is an average value of the absolute values of the dimensional differences multiplied by an appropriate value. A metal can end winding appearance inspection method characterized by the following. 6. The metal can according to any one of claims 1 to 5, wherein the metal can is transported on a conveyor, and a sensor detects that the metal can enters the imaging range of the TV camera. A metal can end-winding appearance inspection method that allows an image to be taken by a camera. 7. In a metal can end winding appearance inspection device,
A ring-type illuminating device arranged above the tightening portion, a television camera arranged coaxially with the center of the illuminating device for imaging a reflection image of the upper end of the tightening by the illumination light, and a digital multi-tone image Means for converting into an image, and means for sampling a ring-shaped image of the upper end of the tightening in the image in the circumferential direction and monitoring the size of each ring width to determine whether the tightening is good or bad. Can end winding appearance inspection device. 8. Converting a ring-shaped image of an end winding portion of a circular can captured by a TV camera into a digital multi-tone image,
This digital ring image is sampled in the circumferential direction,
An image processing method in which the size of the ring width is monitored to determine whether the tightened portion is good or bad.