JPH04370746A - Detecting method for defective appearance of capsule - Google Patents

Abstract

PURPOSE:To enable a failure detection rate at an engagement part of a capsule by determining a defective article when the number of blocks which appear at a projection pattern in a direction along a seam of a differential binary-coded image of the capsule exceeds a specified value. CONSTITUTION:When a color of a cap 01 is different from that of a body 02 and no printing mark is present on both, a projection pattern (consisting of an integration value for each X coordinate of a picture element of the differential binary-coded image) which is formed by projecting a differential binary- coded image near a seam 03 in a direction (Y-axis direction) in a direction along the seam 03 has a certain height in Y-axis direction and is shown as bars (each bar is called a block) with a certain width in X-axis direction. Also, it is shown as a bar graph where a projection of the differential binary-coded image according to crazing 04 which is generated at the cap 01 is added. Then, a height of the projection pattern is cut off at a specified threshold height Yth, thus obtaining the number of blocks Bth which are housed within the height Yth. Then, in the case of, for example Bth >7, it is judged to be a defective conforming capsule.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention is applicable to a capsule appearance inspection device, etc., which detects defects such as defects in the capsule fitting part (seam part), dirt on the surface, and cracks using the external appearance image of a hard capsule obtained by an iTV camera. This invention relates to a defect detection method. Note that in the following figures, the same reference numerals indicate the same or corresponding parts.

0002

[Prior Art] In a capsule appearance inspection device, a video camera is used to detect defects in the appearance of a fitting part (seam part) between a capsule cap and a body. Of the video signals (analog signals representing the shading of the image) obtained by raster scanning the captured image of the capsule by Conditions that a good sample has in terms of length, continuity, degree of constriction, etc. of converted images are set in advance, and the differential binarized image of a defective sample, such as a sample with cracks or chips, satisfies the above conditions. The mainstream method is to determine that the capsule is defective if the capsule is not present. Note that the differential binary image here refers to a video signal obtained by raster scanning an image of a target object using a video camera (
This is an image created using a so-called differential binarization signal, which is obtained by differentiating an analog signal representing the shading of an image and binarizing this differentiated signal using a predetermined threshold (referred to as differential binarization processing). This is an image in which parts where pixel values change (for example, the outline of a target object, a seam, etc.) of a normal grayscale image appear as significant.

0003

[Problems to be Solved by the Invention] However, cracks, which are a type of defect in the fitting part, and insertion of very thin hair into the fitting part, etc., can be prevented by using the differential 2 of the fitting part.
It has little effect on the digitized image and is often difficult to detect defects. Similar trends are observed for other capsules, such as capsules with different colored caps and bodies with printed marks, and capsules with the same color but with and without printed marks. Therefore, the present invention provides capsule appearance defect detection that can improve the detection rate of defects such as defects in the capsule fitting part (seam) and stains near the fitting part, such as the above-mentioned cracks and insertion of hairs. The task is to provide a method.

0004

[Means for Solving the Problems] In order to solve the above-mentioned problems, the defect detection method of claim 1 provides a method for detecting a defect in a container configured by fitting a cap (such as 01) and a body (such as 02) into a container. In a method for detecting an appearance defect near a seam (such as 03) between the cap and the body of a hard capsule encapsulating a block, the block appearing in a projection pattern in a direction along the seam of a differential binary image of the capsule. When the number (such as the number of blocks within a threshold Bth) is greater than a predetermined value, the capsule is determined to be defective,

In the defect detection method according to claim 2, in the defect detection method according to claim 1, the number of blocks is limited to only blocks whose height appearing in the projection pattern is less than or equal to a predetermined value (threshold height Yth, etc.). be the count value of , and

[0006] In the defect detection method according to claim 3, in the defect detection method according to claim 1 or 2, the capsule has caps and bodies of different colors or the same color, and caps and bodies of the same color. This includes those with a printed mark such as a drug number on one or both of them and those without.

[0007]

[Operation] Capsules in which the number of blocks below a predetermined height that appears in a projection pattern in the direction along the seam of a differential binary image near the seam of the capsule exceed a predetermined value are determined to be defective.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. 1 to 4. Figures 1 (A) and (B) show the joint between a good capsule 1A and a cracked defective capsule 1B, in which the cap 01 and body 02 have a different color (different brightness), and there are no letters (print marks) on either 01 or 02. Differential 2 near 03
A projection pattern (that is, created by projecting the valued image in the direction along the seam 03 (in this example, the Y-axis direction) (that is, on the X-axis)
In other words, an example of a pattern consisting of integral values for each x-coordinate of pixels of the differential binarized image will be shown. Here, the projected pattern of a good capsule 1A without a printing mark (Fig. 1A) is a bar (this one A bar graph in which one bar is called a block) is concentrated near the seam 03. In addition, the projection pattern of the cracked defective capsule 1B (Fig. 1(B)) is a bar graph in which the projection of the differential binary image of the crack 04, which occurs from the seam 03 to the ball of the cap 01, is added. There is.

Therefore, in order to detect defective products, the height of each projection pattern (A) and (B) must be set to a predetermined threshold height Y.
Cut at th. In this example, 0<Yth<21 ------(1). Then, the number of blocks that fall within this height Yth (referred to as the number of blocks within the threshold value) Bth is determined. (In other words, blocks cut by the height Yth are naturally ignored as appearing blocks and are not added to the number of blocks Bth.In this example, if Bth>7────────(2) Judging to be a good capsule.

FIGS. 2A and 2B show a non-crackable capsule 2A in which the cap 01 and body 02 are different colors as in FIG. The projection pattern of each differential binarized image of the non-defective capsule 2B onto the X axis is shown. In other words, as shown in FIG. 2, for example, when a print mark appears on the screen that images the capsule during transportation of the capsule (the presence of this print mark is determined by the well-known method (length , continuity, etc.), blocks based on printed marks also occur within the projection pattern. In such a case, the above (1
), the conditions corresponding to equation (2) are changed to determine whether the product is defective.

For example, if the number of blocks based on the print mark is small, the failure judgment standard value for the number of blocks within the threshold Bth is set while maintaining the threshold height larger than the block height of the print mark portion, such as threshold height Yth1 in FIG. It may be added by the number of blocks generated by printing marks, but if there are many blocks based on printing marks, a threshold height that cuts off the height of the blocks based on printing marks, such as threshold height Yth2 in FIG. 2, may be used. is selected, and the number of blocks caused by printing marks is not added to the defect determination reference value, or a small number is added. However, if the printing mark does not appear on the capsule imaging screen due to the transporting posture of the capsule, and therefore a block of printing marks is not recognized in the projection pattern of the differential binarized image, a defective judgment is made under the same conditions as in Figure 1 above. .

FIGS. 3(A) and 3(B) each show cap 0.
Figures 4(A) and 4(B) show the same projection patterns as above for a good capsule 3A and a cracked defective capsule 3B when the body 1 and the body 02 are the same color (same brightness) and have no printed mark. The same projection patterns as above are shown for a non-defective capsule 4A and a cracked defective capsule 4B when the print marks are the same color and the print marks are present on the cap and body.

When the cap 01 and the body 02 are capsules of the same color as shown in FIGS. 3 and 4, no signal change is observed at the seam 03, so special changes such as applying a light diagonally, etc. The lighting method is designed to obtain signals at the seams. As a result, due to shadows appearing on the screen, the number of blocks in the projection pattern increases, unlike in the case of capsules with different colored caps and bodies. However, if there is no printed mark as shown in Figure 3, the value of the block detection threshold Bth is increased as shown in the formula (3) below, and if Bth > 14 (3), it is determined that the capsule is a defective product. I'll do what I do. In addition, if there are printing marks on the cap or body as shown in Figure 4, add the number of blocks of the printing marks as in the case of Figure 2, or cut out the block height of the printing marks using the threshold height Yth. Conditions are set based on the following.

[0014]

According to the present invention, by comparing the number of blocks in a projection pattern obtained by projecting a differential binary image near the fitting (joint) part of the cap and body of the capsule in the direction along the joint, Since the capsule is determined to be defective, the rate of detecting defects such as cracks in the fitting portion, hair insertion, etc. can be greatly improved. FIG. 5 is a diagram comparing the defect detection rate for each defect level of cracked defective products according to the present invention between the conventional defect detection method and the defect detection method of the present invention. As can be seen from this figure, the defect detection rate based on the present invention is approximately twice as high as that of the conventional method.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a projection pattern of a differential binarized image of a capsule without different brightness characters as an example of the present invention.

[Figure 2]
A diagram showing a projection pattern of a differential binarized image of a capsule with characters of different brightness as an embodiment of the present invention

FIG. 3 is a diagram showing a projection pattern of a differential binarized image of a capsule without letters of the same brightness as an example of the present invention.

FIG. 4 is a diagram showing a projection pattern of a differential binary image of a capsule with characters of the same brightness as an example of the present invention.

[Figure 5] A diagram comparing the detection rate of cracked defective capsules based on the present invention with the conventional detection method.

[Explanation of symbols]

01 Cap 02 Body 03 Seam 04 Crack 1A Good capsule with no different lightness text 1B Cracked defective capsule with no different lightness text 2A Good capsule with different lightness text 2B Cracked defective capsule with different lightness text 3A Same lightness without text Good quality capsule 3B
4A cracked defective capsule with same brightness character 4B Good capsule with same brightness character 4B Cracked defective capsule with same brightness character Yth Threshold height Bth Number of blocks within threshold

Claims (3)

[Claims] 1. A method for detecting an appearance defect in the vicinity of a seam between the cap and body of a hard capsule formed by enclosing a drug in a container formed by fitting a cap and a body, wherein: A method for detecting defective capsule appearance, characterized in that when the number of blocks appearing in the projection pattern of the digitized image in the direction along the seam is greater than a predetermined value, the capsule is determined to be defective. 2. The defect detection method according to claim 1, comprising:
The capsule appearance defect detection method is characterized in that the number of blocks is a count value of only blocks whose height appearing in the projection pattern is less than or equal to a predetermined value. 3. The defect detection method according to claim 1 or 2, wherein the capsules include capsules in which the cap and body have different colors, and capsules in which the cap and body have different colors, and capsules in which the cap and the body have different colors, and capsules in which the cap and the body have different colors, and the cap and the body have different colors, and A method for detecting defective appearance of capsules, characterized in that capsules include capsules with and without printed marks such as drug numbers.