JP3022145B2 - Tablet inspection apparatus and tablet inspection method provided with image recognition means - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a tablet inspection apparatus and a tablet inspection method having image recognition means for detecting the number of tablets enclosed in a medicine bag.

[0002]

2. Description of the Related Art FIG. 4 shows an example of such a tablet inspection apparatus.

[0003] This system automates tablet packing work in hospitals and pharmacies, and includes a tablet packing device (1) and an image recognition processing device (3) including a microcomputer. In the tablet packaging device (1), a tablet supply unit (11) in which a plurality of types of tablets are stored by type is provided, and various tablets discharged from the tablet supply unit (11) are fed to a belt conveyor (12). After being transported by (13) and put into the transparent packaging paper (14), it is sealed for each person by the packaging mechanism (17) provided below the belt conveyor (13). Is done.

[0004] Also, in the transport path of the wrapping paper (14), the wrapping paper (1) is provided.
Printing mechanism (not shown) that prints the patient name and usage on the surface of 4)
Is deployed. The packaging paper thus obtained (14)
Is separated for each person to obtain a medicine bag (15) shown in FIG.

[0005] The tablet supply unit (11) and the packaging mechanism (1)
7) is a computer (not shown) that instructs their drive
And the user simply enters a prescription into the computer and the task is performed automatically.

The following image recognition means is provided outside the tablet packaging device (1). The medicine bag (15) discharged from the tablet packaging device (1) and connected in a belt shape is conveyed in a horizontal direction, and a camera (2) for photographing the surface of the medicine bag (15) is provided above the conveyance path. I have. Also, below the transport path,
A lighting device (21) for irradiating the back surface of the medicine bag (15) is provided.

[0007] An image from the camera (2) is input to an image recognition processing device (3) composed of a microcomputer or the like. The image recognition processing device (3) detects pixels as described later and counts the number of tablets (9) enclosed in the medicine bag (15). The image recognition processing device (3) includes a monitor (4)
Are connected, and the image processing process is displayed.

The reason why the lighting device (21) irradiates only the back surface of the medicine bag (15) is that the illumination device (21) also irradiates the medicine bag (15) from above.
This is because the surface of the packaging paper (14) of the medicine bag (15) reflects the illumination, causing a so-called "flash", which makes it impossible to photograph the tablet correctly.

FIG. 6 is a circuit block diagram for executing image recognition processing in the image recognition processing device (3). First, pixels of a video signal obtained by imaging a plurality of tablets are input to the inverting input terminal of the comparator (50). To the non-inverting input terminal of the comparator (50), a threshold value having a substantially middle density between the tablet density and the background density is input, and the comparator (50) performs a so-called binarization process. Binarization processing means that among pixels of a grayscale image, pixels having a density equal to or lower than a threshold value are set to “1”,
In the process of setting a pixel having a density equal to or higher than the threshold value to “0”, for example, Japanese Patent Publication No. Hei.
No. 83 is a known technique.

When the illuminator (21) irradiates the tablet (9) from below, the image taken by the camera (2) includes the portion shaded by the tablet or the like shown by the hatched portion in FIG. Are divided into portions that have passed through the medicine bag (15). By the comparator (50)
The image is binarized as shown in FIG. That is, contrast conversion is performed on the original image, and the outline of the tablet (9) is emphasized. In the binary image, noise (18) caused by characters printed on the packaging paper (14) remains.

Then, noise (18) contained in the binary image
Is removed by the contraction / expansion processing circuit (52).

The shrinking process is as follows. FIG. 8A shows an example of an image state before the contraction processing, and FIG. 8B shows a pixel state after the contraction processing. In FIGS. 8A and 8B, a point where a pixel is “0” is indicated by a dot.

First, the screen is divided into pixel units, and each pixel on the screen is detected. Then, pixels located above, below, left and right of the point where the pixel is “0” are erased from the screen.

For example, in FIG. 8A, since the pixel in the second row and the second column is “1”, in FIG.
The upper, lower, left and right pixels of the pixel data in the second row in the row disappear.

In the dilation process, a pixel is generated when there is no pixel at a point located on the upper, lower, left, or right side of an arbitrary pixel, contrary to the erosion process. FIG. 8C shows a pixel state after dilation processing of the pixel state shown in FIG. 8B (refer to “Basic Techniques for Image Processing” (published by Technical Review Company), pp. 54 to 56). By performing such contraction / expansion processing several times, the noise (18) is removed.

After the noise (18) is removed, the contact detection circuit
At (53), the two objects of the binary image are in contact with each other, or contact points located at both ends of the area where they cross each other, for example, a
A pair of contacts indicated by points b and b are detected one after another. Next, the contact connection circuit (54) connects the pair of contacts to create a boundary line ab indicated by a broken line. As shown in FIG. 9B, after the images of the two objects touching each other are separated by a boundary line by the separation circuit (55), the images are subjected to labeling processing by the count processing circuit (56) and included in the images. Count the number of connected components.

Here, the labeling process means that a pixel exists at a point adjacent to an arbitrary pixel in the vertical and horizontal directions.
This is a process in which those pixel groups are regarded as one group and labels are assigned. That is, the number of groups having the same label is measured. FIG. 12 shows a pixel state after the four-link labeling processing. Pixel groups surrounded by a dashed line are divided into individual groups, and labels are assigned to the respective groups (see “Basic Techniques for Image Processing”, pp. 45 to 49).

In the inspection of the tablet (9) in the medicine bag (15),
Tablets often overlap each other. Even in such a case, the above-described method can separate the overlap between the tablets and detect the correct number.

[0019]

There are various types of tablets, including translucent tablets. For example,
Alfalol, epanol, and flupanon software are translucent capsules (see “Drug Dictionary: Pill Book 93-94”, published by Yakuhin Tohokusha). If these translucent tablets are mixed with general tablets, the above-described image recognition processing procedure cannot be performed correctly. The reason is shown below.

FIG. 2 shows the concentration distribution when a translucent tablet and a general tablet are irradiated from below with a lighting device (21).
In the original image of the medicine bag (15) by the lighting device (21), the density of the portion where the translucent tablet exists is higher than the density of the portion where the general tablet exists.

That is, when binarizing an image of a medicine bag (15) in which translucent tablets and general tablets are mixed, if the threshold value is not set high, the translucent tablets will not be binarized. By
It is set to “0”.

However, as described above, noise remains in the binary image due to the characters and the like printed on the packaging paper (14).
In this case, when the threshold value is set to a high value, with the normal threshold value, minute dust and wrinkles on the removed packing paper (14) are also detected as noise, and "1" is obtained by the binarization process. " As a result, the number of connected pixels in the character portion increases, and noise cannot be removed in the contraction / expansion processing. Therefore, the correct number of tablets cannot be detected.

As a method of the binarization processing, a so-called adaptive type in which an original image is divided into several small areas, a histogram of pixel density data is obtained for each area, and a threshold value is set for each area. A threshold method has also been proposed (see "Nikkei Mechanical", July 30, 1984, pages 205 to 206).

However, even with this method, the number of translucent tablets cannot be accurately detected. This is because, as shown in FIG. 10 (a), in the region where the outer periphery of the general tablet is in contact with the outer periphery of the translucent tablet, where the ratio of the general tablet and the translucent tablet is almost equal, as shown in FIG. The number of pixels constituting the shape of the tablet is substantially equal to the number of pixels constituting the shape of the translucent tablet.

Therefore, in the histogram shown in FIG. 11 in which the density data of the pixels are plotted on the horizontal axis and the number of pixels is plotted on the vertical axis, in order to distinguish the general tablet from the background, the threshold value is
It is approximately the middle value between the concentration of a general tablet and that of a translucent tablet.
In this state, the translucent tablet is detected as having the same density as the background, and disappears from the binarized image. Therefore, in the binarized image, the outline of the translucent tablet is determined as shown by the thick solid line in FIG. 10B, and is not configured with the correct curve or outline, and the element separation processing may not be performed accurately. is there.

An object of the present invention is to clarify a tablet inspection apparatus and method capable of correctly counting the number of tablets even when a translucent tablet is contained in a medicine bag.

[0027]

The tablet inspection apparatus compares the density of the pixel data input from the imaging means with a threshold value having a sufficiently low density level that can identify the density of a general tablet from the background density. A general tablet recognition processing means for detecting pixels of the general tablet (9), processing the pixels and counting the number of general tablets in the medicine bag (15), and a limit of the density distribution of the translucent tablet. Two thresholds, a lower threshold and an upper threshold, are set, and the density level of the pixel input from the imaging means is set.
Translucent tablet with a value between two thresholds from the bell
A translucent tablet recognition processing means for detecting the number of translucent tablets in the medicine bag (15) by processing the pixels, by adding the number data of the tablet outputted from respective their
And a calculation control means for outputting the number of all the tablets in the medicine bag obtained in the above.

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings.

The mechanism in the tablet packaging device (1) and the connection between the image recognition processing device (3) and the monitor (4) are the same as in the prior art, so that the description is omitted.

FIG. 1 is a circuit block diagram of the image recognition processing device (3). The image recognition processing device (3) includes a general tablet recognition processing unit (5) and a translucent tablet recognition processing unit (6) inside. Input terminal (30) to which the image signal of camera (2) is input
Is connected to the input side of both processing units (5) and (6).
The output side of (6) is connected to the arithmetic control circuit (7). The output side of the arithmetic control circuit (7) is connected to the output terminal (31) of the image recognition processing device (3).

As shown in FIG. 3, the configuration of the general tablet recognition processing section (5) is the same as that of a conventional image processing recognition apparatus. That is, the general tablet recognition processing unit (5) performs quantity inspection of general tablets, and the threshold value given to the non-inverting input terminal of the comparator (50) can distinguish only background and general tablets. , Set a sufficiently low value. As a result, the image of the translucent tablet does not appear in the image signal output from the general tablet recognition processing unit (5), and the inspection of the medicine bag (15) in which the general tablet and the translucent tablet are mixed is performed. However, only the quantity of general tablets is detected. From the contraction / expansion circuit (52) to the count processing circuit (5
Up to 6), the processing is the same as the conventional processing shown in FIG. 6, and a description thereof will be omitted.

In the translucent tablet recognition processing unit (6), as shown in FIG. 3, a window comparator (8) having two comparators (80) and (81) and an AND gate (82) is provided with a contraction / expansion circuit. (62) is connected. The output terminals of the comparators (80) and (81) are connected to the input terminals of the AND gate (82). The contraction / expansion processing circuit (62) includes a contact detection circuit (63), a contact connection circuit (64), and a separation circuit as in the general tablet recognition processing section (5).
(65), the count processing circuit (66) is connected in order. The processing from the contraction / expansion processing circuit (62) to the count processing circuit (66) is the same as the processing from the contraction / expansion circuit (52) to the count processing circuit (56) of the general tablet recognition processing section (5). .

The lower threshold and the upper threshold shown in FIG. 2 are input to the window comparator (8). The lower threshold and the upper threshold indicate the lower limit and the upper limit, respectively, in the density distribution of the translucent tablet. Show.

The video signal is input to the non-inverting input terminal and the lower threshold value is input to the inverting input terminal to one comparator (80) of the window comparator (8). Therefore, when the density of the pixel of the video signal is higher than the lower threshold, "1" is output from the comparator (80), and when it is lower than the lower threshold, "0" is output. .

The other comparator (81) receives the upper threshold value at the non-inverting input terminal and the lower threshold value at the inverting input terminal. When the density of the pixel of the video signal is lower than the upper threshold, "1" is output from the comparator (81), and when it is higher than the upper threshold, "0" is output.

As is well known, the AND gate 82 does not output "1" unless both outputs from the comparators 80 and 81 are "1". That is, the window comparator (8) outputs "1" only when the density of the pixel of the video signal is between the two threshold values.

For example, when a pixel of a general tablet is input to the window comparator (8), each of the comparators (80) (8)
Since "0" is output from 1), the translucent tablet recognition processing unit (6) does not detect the pixels of the general tablet.
In this way, only pixels of the video signal of the translucent tablet are detected.

The output from the window comparator (8) is subjected to the contraction / expansion processing by the contraction / expansion circuit (62) as described above, and the contact detection circuit (63) detects the contact point. The contacts are connected by (64), and are separated for each element by the separation circuit (65). Count processing circuit
According to (66), after performing the labeling process, the number of translucent tablets is detected, and the quantity data of the translucent tablets is output to the arithmetic and control circuit (7).

The arithmetic control circuit (7) receives the quantity data and the video signal from the general tablet recognition processing section (5) in addition to the quantity data of the translucent tablet, and adds both quantity data. When the added data is different from the number of tablets that should be originally enclosed in the medicine bag (15), the user is notified of this by, for example, emitting a warning sound.

The video signal output from the arithmetic and control circuit (7) is input to the monitor (4) via the output terminal (31).

The tablet inspection device is a general tablet recognition processing unit (5)
And a translucent tablet recognition processing unit (6) to detect the number of general tablets and the number of translucent tablets separately, add both the detection data, and detect the tablets in the medicine bag (15). ing.
Therefore, even when a translucent tablet is included in the medicine bag (15), even if the translucent tablet and the general tablet partially overlap, the number of tablets can be accurately detected.

It is assumed that the two threshold values input to the window comparator (8) are set in advance.
Before the inspection, the density data of the translucent tablet to be inspected may be obtained and input to the device when the device is used.

The description of the above embodiments is for the purpose of illustrating the present invention, and should not be construed as limiting the invention described in the claims or reducing the scope thereof. Further, the configuration of each part of the present invention is not limited to the above-described embodiment, and it is needless to say that various modifications can be made within the technical scope described in the claims.

[0044]

[Operation and effect] The tablet inspection device detects pixels of a general tablet.
In addition to the general tablet recognition processing means
It has a means of management. The translucent tablet recognition processing means includes:
Upper and lower thresholds, which are the concentration distribution limits of transparent tablets
Are set. Everything in the medicine bag
When counting the number of tablets, the translucent tablet recognition processing
Stage detected pixel with value between two thresholds
Then, the pixel is processed, and the number of translucent tablets in the medicine bag (15)
Is detected. Thereafter, the sum of the number of the general tablets detected by the general tablet recognition processing means and the number of the general tablets is calculated by the arithmetic control means. Therefore, even if the translucent tablet is enclosed in the medicine bag (15) together with the general tablet, the number of tablets can be accurately detected.

[Brief description of the drawings]

FIG. 1 is a circuit block diagram of a tablet inspection device.

FIG. 2 is a graph showing the relationship between the density value of a pixel and the number of pixels having each density value.

FIG. 3 is a block diagram showing the insides of a general tablet recognition processing unit and a translucent tablet recognition processing unit.

FIG. 4 is a perspective view showing a configuration of the present invention and a conventional tablet inspection apparatus.

FIG. 5 is a perspective view of a medicine bag.

FIG. 6 is a circuit block diagram of a conventional image recognition processing device.

7A is an image before binarization processing, and FIG. 7B is an image after binarization processing.

8A is a diagram showing an image before a contraction / expansion process, FIG. 8B is a diagram showing an image after a contraction process, and FIG. 8C is a diagram showing an image after a dilation process.

9A is a diagram showing an image after contact detection and connection of contacts, and FIG. 9B is a diagram showing an image after separation of each element.

FIG. 10 (a) shows 1 when the adaptive threshold method is used.
FIG. 3B is a diagram showing a state in which a general tablet and a translucent tablet are in contact with each other in two regions, and FIG.

11 is a graph showing the relationship between the density value of a pixel and the number of pixels having each density value in one region in FIG.

FIG. 12 is a diagram illustrating a state of a pixel after a labeling process.

[Explanation of symbols]

 (9) Tablet (15) Medicine bag

Continuation of the front page (56) References JP-A-3-1499051 (JP, A) JP-A-63-147279 (JP, A) JP2-4948 (JP, B2) JP1-30183 (JP) , B2) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06M 7/00 301 A61J 3/00 310 G06M 11/00 G06T 7/00

Claims (2)

(57) [Claims] 1. An imaging means for photographing a light-transmitting medicine bag (15) in which one or more tablets (9) are enclosed, and a concentration level of a general tablet is set to a sufficiently low concentration level that can be distinguished from a background concentration. The threshold value is compared with the density level of the pixel input from the imaging means, the pixel of the general tablet (9) is detected, the pixel is processed, and the number of general tablets in the medicine bag (15) is measured. in tablets inspection apparatus having a general tablet recognition processing means, to set the two thresholds consisting of lower threshold and upper threshold is the limit of the density distribution of the semi-transparent tablet, or the imaging unit
From the density level of the pixel input from
Translucent tablet recognition processing means for detecting a pixel of the translucent tablet having a value between and processing the pixel to detect the number of translucent tablets in the medicine bag (15); and a translucent tablet recognition processing means. , by adding the number data of the tablet that is output from each of the general tablet recognition processing means, the
A tablet inspection device comprising: an arithmetic control unit that outputs the number of all tablets in the obtained medicine bag . 2. The concentration of a general tablet can be identified from the background concentration.
The threshold value, which is a sufficiently low density level, is compared with the density level of the pixel input from the imaging means, and after detecting the pixel of the tablet, the pixel is processed, and the pixel in the medicine bag (15) is processed. In addition to outputting the number data of the tablets, two thresholds, which are the concentration distribution limits of the translucent tablet, are set, and the two thresholds are set.
Value and the density level of the pixel input from the imaging means.
Translucent tablet having a value between two thresholds
After detecting the pixel of the drug, the pixel is processed, the number data of the translucent tablet in the medicine bag (15) is also output, and both number data are added to calculate the number of all tablets in the medicine bag (15). Tablet inspection method to measure the.