JPH03260876A - Method for detecting mixing of different kind of printed cap - Google Patents

Abstract

PURPOSE:To securely detect the mixing of different kind of printed caps with simple operations by picking-up the image of the can cap to be inspected, transforming this three-dimensonal picture data to two-dimensional picture data, generating sample cap data from this two-dimensional picture data and comparing this sample cap data with the data of a regular can cap obtained in advance. CONSTITUTION:For the picture signal of a can cap 7 inputted from a CCD color camera 2, RBG separation is executed and the signal is fetched into a color frame memory 6 as the three-dimensional picture data. Then, RGB-XY transformation is carried out for this data by a personal computer 5 and after various arithmetic processings, the detection is performed for different kinds of printed cap. Namely, the sample cap data is generated by excluding an area, which is overlapped to the two-dimensional picture data of not-printed can cap calculated in advance, from the two-dimensional picture data and this sample cap data is compared with the sample cap data of the regular can cap calculated in advance. Therefore, the different kind of printed caps to be mixed in the regular can cap is discriminated. Thus, the different kinds of printed cap can be easily discriminated with high accuracy.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a method for detecting contamination of different types of printed lids, and particularly relates to a method for detecting contamination of different types of printed lids that uses optical detection means to detect different types of printed lids. .

[Conventional technology]

BACKGROUND ART In recent years, characters, patterns, etc. have been printed and displayed on the surface of can lids in predetermined colors (can lids with this printing are hereinafter referred to as "middle@lids"). The color of the characters, etc. displayed on the lid generally differs depending on the type of contents. For example, on a printed lid containing juice, the color of the letters etc. may be red, and on a printed lid containing coffee, the color may be blue.

The printed lid and can body are manufactured on separate lines, and the printed lid is wrapped around the can body in the final can manufacturing process to complete the can. At this time, since printing is also applied to the can body separately from the markings, it is necessary to unify the contents of the printing applied to the can body and the printing applied to the printed lid. Using the above example, it is necessary to wrap a printed lid printed in blue around a can body printed with information related to coffee.

However, in cases where the cans being manufactured are changed from juice cans to coffee cans, although this is extremely rare, if the lid production line is marked with the red color for juice after the above change. A printed lid (such a printed lid is referred to as a different type of printed lid) may remain, and in this case, a disadvantage arises in that the printed juice lid is wrapped around the coffee can body.

Therefore, a means for detecting the mixing of different types of printed lids is provided in the production line marked *ii to prevent the above-mentioned inconvenience from occurring.

The conventional method for detecting this type of foreign material contamination is to take an image of the printed lid with a camera, and then use a combination
- A method has been adopted in which different types of printed lids are discriminated by g*b of the printed content by processing the lid. Specifically, first, the image captured by the camera is divided into red (R), green (G), and blue (B).
It is plotted on the three-dimensional coordinates shown in FIG. 20 using the three primary colors (luminance can be used instead of green) as parameters. At this time, the number of plot points plotted on the coordinates is determined by the number of pixels of the camera and the area of the mark w4I.

FIG. 21 shows an example of a printed lid taken with a camera and all 70 points removed.

Further, a reference range in which the plot points can be taken in the case of a regular printed lid is set in advance in the computer. In the case of the same figure, the circumference that can be taken by a regular printed lid is red (R).
The range is from 1 to R2, and green (G) is 1 from G+ to G2.
and blue (B) is the range from B1 to B2.

Therefore, if there are many plot points within the rectangular parallelepiped portion shown by the solid line in the figure, the printing cover is a regular mark Il!
It can be determined that

Conversely, if there are many plot points outside the rectangular parallelepiped portion, it is determined that the printed lid is a different type of printed lid, thereby detecting tI4I in the different type of lid.

[Problem to be solved by the invention]

However, the above-mentioned conventional method for detecting the contamination of ill
, Stomach (B), 1) (G), Red (R) fX, Y.

Since the configuration was such that colors were expressed using three parameters corresponding to each axis of Z, there were the following problems.

First, when setting the standard II of the regular printing lid on the computer, you must set the upper and lower silver values for each of the three primary colors while looking at the extracted colors that appear on the combination CRT (cathode ray tube). It didn't happen. This setting operation was a manual operation, and a total of six points had to be adjusted, so there was a problem that setting the reference range was very troublesome.

Furthermore, during detection, the reference range is rectangular, so the detection accuracy is low, and it is difficult for an IN- that receives a printed lid printed with a color that is similar to the regular printed lid to distinguish it. II can't do it! There was Ii.

The present invention has been made with the above points in mind! 1. It is an object of the present invention to provide a method for detecting contamination of different types of printed lids, which can reliably detect contamination of III of different types with a single operation.

[Means to solve the problem]

In order to solve the above problems, in the method for detecting contamination of different types of printed lids according to the present invention, the can lid to be inspected is imaged by an imaging means, and this ms
The three-dimensional image data composed of the three primary colors of the can lid generated by the means is mapped onto a two-dimensional plane and converted into two-dimensional image data, and then, from the three-dimensional image data, a plain color obtained in advance is Sample lid data is generated by deleting the area that overlaps with the two-dimensional image data of the can lid (plain lid data), and this sample lid data is converted into sample lid data of a regular can lid obtained in advance (regular lid data). The feature is that it is possible to distinguish between a regular can lid and a different type of printed lid by comparing the can lid with the original can lid.

(Function) In the above method for detecting contamination among different types, since three-dimensional image data is converted into two-dimensional data, handling of the data becomes easy. Therefore, the operation for setting the regular lid data is easy, and since the sample lid data can be compared with the regular lid data on the plane coordinates, this data can be easily compared, and the detection accuracy can be further improved. can be improved.

In addition, the sample lid data is the image of the entire lid! This data is obtained by deleting the plain lid data from the two-dimensional JII image data with information. Therefore, the sample lid data includes only the characters embossed, and the judgment is made by comparing the data, so that the detection accuracy can also be improved.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings. First, using FIG. 18, we will explain a simulation device used in a method for detecting contamination of different types of bleached lids, which is an embodiment of the present invention! 1 will be explained.

The simulation equipment f1 includes a solid-state III element color television 2 (hereinafter referred to as a COD color camera), a light magnification circle 1j3. The optical system includes a strobe lamp house 4 and the like, a personal computer 5, a color frame memory 6, and the like.

Note that 7 in the figure indicates the can lid to be inspected. It is assumed that the upper surface of the can I7 is inscribed with letters, pieces, etc. in a predetermined color.

A strobe lamp is disposed within the strobe lamp house 4, and the strobe lamp is configured to be supplied with power from a strobe power source controlled by a personal computer 5. When the strobe lamp is turned on, the light is guided to the ring light guide 9 through the optical fiber 8. The light emitted from this ring light guide 9 is
The light is made uniform within the light magnification circle 113 and then irradiated onto the top surface of the can lid 7. Thereby, halation of the can lid 7 can be reduced. In addition, to capture images, lens 10 (f-25 agm) with a 2wm extension ring for close-up shooting, R (red), G (1),
B (blue) min 1 m (DCCD force 5 - camera 2 used.

Image of can I7 input from CCD color camera 2! The 11 signals are separated into RBG and taken into the color frame memory 6 as three-dimensional image data.
The internal storage device after converting it to pit data (
memory). Memory is 6 for each RGB
4 has the pit capacity. In addition, CCD color camera 2
11! The resolution of the i-plane is 256 dots in both the vertical and horizontal directions.

Cans stored in color frame memory 6! The image data of 7 is converted into RGB-
XY conversion is performed, and after various arithmetic processing, detection of different types of seals and lids is performed.

Next, the principle of the inspection method performed using the simulation device 11 will be explained. In the inspection method of the present invention, it is necessary to create plain lid data and regular lid data in advance before starting the inspection. First, how to obtain the lllll11 lid data will be explained.

A plain lid refers to a can lid that has not been marked.

To obtain plain lid data, use the 18th Ii! A plain lid is set in place of the can lid 7 shown in I, and the CCD color camera 2 images this. The image signal of the plain lid inputted from the CCD color camera 2 is applied to a three-dimensional R18,6.
Therefore, the image is subjected to IIR using three parameters and taken into the color frame memory 6 as three-dimensional image data.

The three-dimensional ij* data of the plain lid captured in the color frame memory 6 is supplied to the personal computer 5. 3 captured in 3D on personal computer 5
The dimensional image data is converted into a 2-dimensional plane and converted into 2-dimensional image data constituted by two parameters (hereinafter, this image conversion will be referred to as RGB-XY conversion).

A specific conversion method of RGB-XY conversion will be explained using the first example. The figure shows an RGB color system in which R, G, and B are arranged in three orthogonal wheels. Assume now that a certain part (corresponding to one dot) of the j-color image captured by the RGB color camera 2 has the color at the position of the point Po shown in the figure.

RGB-XY conversion: Create a plane that cuts each ring of R, G, and B, and define this as the XY plane. Then, a straight line is drawn from the point P0 to the origin O, and the point where this straight line crosses the above-mentioned XYsFflj is the first figure in Patent 2 which features snow! ! FIG. 1 is a front view of the XY plane shown in FIG. If we take the XY ring with the center of this XY plane as the origin, the above point P+ is X+
, Y+ (image data expressed using two parameters as described above is referred to as two-dimensional image data). That is, RGB
According to the −xY transformation, the point Pa of the three-dimensional beast consisting of R, G, and B! By mapping to the XY plane, the color of point P0 can be expressed using two parameters. This allows images and data to be compressed, making it easier to handle and process image data. Furthermore, since X and Y show the same value as long as the ratios of R, G, and B do not change, they have the characteristic that they are not easily affected by changes in the amount of light from the light source. In addition, what is shown with a broken line in FIG. 2 is RlG in FIG.
, B.

Using the RGB-XY conversion described above, the three-dimensional image data captured in three dimensions of the plain lid is converted into two-dimensional image data. The image captured by the CCD color camera 2 undergoes RGB-XY conversion for all pixels, and
Plotted on the Y plane. Now, if the color of the plain lid is the metallic color of aluminum (substantially white), the two-dimensional image data will be as shown in FIG. This plain lid 2
The dimensional image data is stored in memory within the color frame memory 6.

Next, how to obtain the regular lid data will be explained.

Positive I! The term "lid" refers to the above-mentioned plain lid with a predetermined regular printing applied thereto. In order to obtain the regular lid data, first, the regular lid is imaged with the CCD color camera 2 to obtain three-dimensional image data, which is then subjected to RGB-XY conversion. Assuming that the printed color is red, the contents of the two-dimensional image data will be as shown in FIG. 4, for example. The data shown in FIG. 4 is data content in which both the plain portion and the printed portion of the regular lid are extracted.

Next, the two-dimensional image data of the regular lid obtained as described above is compared with the plain lid data obtained by the method described above, and the common portion is divided by m from the two-dimensional image data of the regular lid. This process will be explained using FIG. 5.

In the figure, the part surrounded by a solid line is the range in which the color components of the regular lid are distributed, and the part surrounded by the broken line is all in which the color components of the plain lid are distributed. Therefore, both have common parts shown in matte finish. This common portion corresponds to a portion of the regular lid that is not printed and is in a plain state. Therefore, by removing the plain portion from the normal*ii two-dimensional image data, the data from which the plain portion has been removed becomes data showing only the printed portion. The two-dimensional image data generated in this manner and corresponding to only a portion of each regular mark is referred to as regular lid data. This regular lid data is stored in the color frame memory 6.

In the method of the present invention, the detection of contamination with different types of printed lids is performed based on the plain lid data and regular lid data obtained as described above. A method for detecting contamination will be described below.

To detect contamination, first, a sample lid to be inspected is imaged with a CCD color camera 2 to obtain three-dimensional image data, which is then subjected to RGB-XY conversion to generate two-dimensional image data. Assume that the result is shown in, for example, No. 6f1. Next, the two-dimensional image data of this sample lid is compared with the plain lid data obtained in advance, and the common parts (portions shown in matte finish in Figure 7) are compared with the two-dimensional image data of the sample lid.
Removed from dimensional image data. It was asked like this,
The data in the area indicated by the solid line in FIG. 8 is referred to as sample lid data.

When the sample lid data is obtained, the previously obtained positive [1 data is taken in from the color frame 6,
Compare each data. Then, the common part of both data, in other words, the area occupied by both data on the XY plane is ii
*If the area of the overlapping portion is large, it is determined that the sample lid is a regular lid, and if the area of the overlapping portion is small, it is determined that the sample lid is a foreign mark i1m. In the above example, as shown in FIG. 9, the common portion (indicated by hatching) between the regular lid data and the sample lid data is small, so it can be determined that the sample lid is a different type of printed lid. In the following explanation, a portion of the sample lid data that is not common to the regular lid data, that is, a portion shown in matte finish in the figure will be referred to as an area of other colors.

As described above, in the method of the present invention, it is possible to determine different types of printed lids by obtaining sample lid data, comparing it with regular lid data, and simply determining whether the common portion of both data is large or small. can.

Further, each data is easy-to-handle data composed of two parameters converted into RGB-XY*. Further, both the sample lid data and the regular lid data are data that reflects only the printed portion in which the area occupied by the plain lid data has been deleted. Therefore, it is possible to easily and accurately detect the mixing of different types of printed lids. Further, the regular lid data that serves as a reference for determining different types of printed lids is obtained through computer processing and is automatically set. For this reason, as in the past, the regular lid standard II! This eliminates the need to manually input the range into the computer, making the detection operation easier. However, since the regular lid data may also include some errors, the operator needs to input the intersections based on these errors.

Next, an example of a specific detection method for detecting the contamination of different types of printed lids performed by the present inventor will be described.

Note that the RGB-XY conversion adopted an (address)-(data) conversion method using a parameter table. When converting the 6-pit data for each RGB input from the RGB color camera 2, the range of data is ±45X on the XY plane.
．． Y is in the range of ±51. However, in the following explanation, the range of X and Y is limited to ±8 for convenience. In addition, as described below, the final result is the color printed on the regular lid (hereinafter referred to as
The areas of the standard color), plain color, and other colors are expressed by the number of pixels (dots) each occupies. In addition, in the multiwaves shown below, locations where there is no data are marked with a "." mark, and the multivalues of X and Y are shown as -8 to 8 in the upper part of the figure and in the upper part of the figure.

Below is the plain lid data and positive I! The procedure for obtaining lid data will be explained.

First, the plain lid data and the main I in the color frame memory 6
! The memory storing the lid data is cleared and the color frame memory 6 is initialized.

Next, five plain lids are sampled by the simulation device M1, and RGB-XY conversion is performed. Based on this conversion result, the number of data corresponding to each coordinate on the XY plane is recorded. Assume that the result is shown in the 10th WU, for example.

Next, five regular lids are sampled, subjected to RGB-XY conversion, and recorded as #1 on the XY plane. Assume that the result is shown in FIG. 11, for example.

The result is stored in the Bu color frame memory 6.

Next, the contents of the regular lid data shown in FIG. 11 and the plain lid data shown in FIG. 10 are compared to determine the number of overlapping pixels. In this example, the overlapping part is the 12th
It is the part surrounded by a solid line in the figure, and its area (number of dots)
becomes 1024 dots. This area is the area of the unprinted plain portion of the regular MM.

Next, of the regular lid data shown in FIG. 11, the 1024 dot portion that overlaps with the plain lid data is deleted from the regular lid data. In Fig. 13, the portion divided by ill is rXJ
It is shown in The sample lid data obtained in this manner is stored in a memory within the color frame memory 6. At this time, the total number of pixels stored in the memory becomes the area of the standard color of the regular lid. In this example, the area of the reference color is 185 dots as shown in FIG.

With the above operations, the setting of 11 [@ lid data and regular lid data is completed. After this, we will import jI & of various sample lids, and Il of different types! A process of determining whether or not is performed will be performed.

Now, the result of sampling the sample lid and performing RGB-XY conversion is, for example, No. 141i! Suppose that it is as shown in I. This two-dimensional image data of sample l and the 10th
Compare the plain lid data shown in WI and find the common parts. The common part is the 15th rj! This is the part surrounded by a solid line J, and its number of pixels is 1062 dots. Generate sample lid data by removing the common parts. The IIi castle of this sample lid data is the part enclosed by the broken line in 1159).

Next, the turbidity amount data stored in the color frame memory 6 is taken in, and the number of pixels in the common portion of the regular lid data (shown at 1131I) and the sample lid data is determined. In the case of this example, what both data have in common is the 1681st area surrounded by a solid line, and the number of pixels is 19 dots. This jlI number of 19 dots becomes the reference color area of the sample lid. Also, among the sample lid data, plain lid data and positive 1! If there is a portion that is not common to any of the data, the total number of pixels of the portion that is not common is the area of the other color. The area of other colors is the area surrounded by solid lines in FIG. 17, and the Ijll (number of jl) of the other colors is 728 dots in this example.

These results are summarized in the table below.

(Unit: dots) From the same table, the plain area of the sample lid is close to the plain area of the regular lid, but the areas of the standard color and other colors are significantly different. Therefore, it can be determined that the sample lid mentioned in this example is a different type of printed lid.

In the embodiments described above, the personal computer 5 is used to perform a routine process according to a predetermined program set in advance to discriminate between different types of Ill. However, this routine process is slow and cannot be used on an actual can lid manufacturing line (about 10 can lids are manufactured every 2 seconds).
It is also possible to adopt a configuration in which the above-mentioned detection of the mixing of different types of printed lids is performed by the hardware shown in FIG.

Roughly speaking, the device 12 (hereinafter simply referred to as a seal) for detecting the mixture of different types of printed lids shown in the same figure consists of a synchronization circuit 13. 0 configured by a data generation circuit 140 and a discrimination circuit 15.
It is composed of an image switching circuit 13 & t, a synchronization signal @ circuit 161, an image-address conversion circuit 17, a window memory 18, etc., and supplies vertical and horizontal synchronization signals to the CCD color camera 2. By adding 7 dresses to each pixel of the screen to be imaged, the image *
-7 Performs the function of performing dress conversion. In addition, by supplying a can signal (Phs) from the can manufacturing line, when a number of can lids pass through the CCD color camera 2, the strobe is turned on. By opening the window memory 18, data outside the can IEX is cut out.

Data generation path 14 is an analog-to-digital converter (A/
D conversion I! >19-21. RGB-XY conversion circuit 22.
It is composed of a plain color/reference color memory 23 and the like. This data generation path 14 outputs three-dimensional image data of the three primary colors RGB of the can lid captured by the CCD color camera 2, respectively.
/D converters 19 to 21 convert the signals into digital signals, which are then subjected to RGB-XY conversion by an RGB-XY conversion circuit 22.

The two-dimensional image data subjected to RGB-XY conversion is supplied to the plain/reference color memory 23, where each pixel is divided into solid color/reference color and stored in the memory corresponding to each pixel. The plain lid data and regular lid data generated in this way are stored in this solid color/reference color memory 23. Note that the sample lid data is not stored in this plain color/reference color memory 23, but is configured to be passed through every drop.

The discrimination path 15 is composed of an area counter circuit 249 and an area comparison circuit 25. The area counter 0 path 24 is the plain cover data supplied from the plain color/reference color memory 23, positive 1! By counting the number of pixels of each data from the data and sample lid data, no jl! Area of region lII,
The chain of each area obtained by the 1-area counter circuit 24 which obtains the area of the reference color region and the area of other color regions is supplied to an area comparison path 25, which calculates the area of the plain color region and the area of the reference color region.

By comparing the areas of other color regions with the regular lid data, it is determined whether or not the lid is of a different type.9 Then, if it is determined that the lid is of a different type as a result of the determination, a signal is supplied to the output circuit 26. , the output circuit 26 supplies, for example, a signal to the manufacturing line to remove the foreign mark-lid.

In the apparatus 12 configured as described above, since different types of printed lids are discriminated by hardware, the processing speed is increased, and the apparatus 12 can sufficiently cope with can manufacturing lines having high line speeds.

〔Effect of the invention〕

As described above, according to the present invention, since the three-dimensional image data captured by the 111 means is converted into two-dimensional image data, data handling becomes easy and different types of printed lids can be easily and highly accurately discriminated. It is easy to set the regular lid data, and since the plain lid data area is removed from both the regular lid data and the sample lid data, discrimination can be made by comparing only the printed parts.
This also has the advantage of improving detection accuracy.

[Brief explanation of drawings]

Figure 1 is a diagram for explaining RGB-XY conversion, Figure 2
J is a diagram showing an XY plane showing RGB-XY converted data, Figure 3 is a diagram for explaining plain lid data, and Figure 4 is a diagram for explaining plain lid data.
5 and 5 are diagrams for explaining how to obtain regular lid data, Figures 6 to 9 are diagrams for explaining the procedure for determining a different type of seal lid from sample lid data, and Figure 10 FIGS. 17 to 17 are diagrams for explaining a specific example of a method for detecting contamination of different types of printed lids, which is an embodiment of the method of the present invention, according to the procedure, FIG. 18 is a diagram showing a simulation device used in the method of the present invention, and FIG. 20 is a diagram showing the color parameters of the RGB color system, and FIG. 21 is a diagram for explaining an example of a conventional method for detecting the contamination of lids with different types of printing. . DESCRIPTION OF SYMBOLS 1... Simulation device, 2... CCD color camera, 5... Personal computer, 6... Color frame memory, 7... Can lid, 13... Image dosing circuit, 14-... Data generation circuit, 15--discrimination circuit.

Claims (1)

[Claims] A can lid to be inspected is imaged by an imaging means, and three-dimensional image data composed of the three primary colors of the can lid generated by the imaging means is mapped onto a two-dimensional plane to generate two-dimensional image data. Next, from the two-dimensional image data, an area overlapping with pre-obtained two-dimensional image data of a plain can lid (plain lid data) is deleted to generate sample lid data, and the sample lid is converted into sample lid data. A method for detecting contamination of different types of printed lids, characterized in that a different type of printed lid that is different from a regular can lid is determined by comparing the data with sample lid data (regular lid data) of a regular can lid obtained in advance. .