JP7280361B2 - Drug identification device, drug identification method and program - Google Patents

Description

TECHNICAL FIELD The present invention relates to a drug identification device, a drug identification method and a program, and more particularly to a drug identification device, a drug identification method and a program for identifying drugs having similar drugs.

2. Description of the Related Art Conventionally, in order to inspect whether or not a drug is dispensed according to a prescription, a technique of photographing the dispensed drug and identifying the drug based on the photographed image has been used.

For example, Patent Literature 1 discloses a technique for inspecting whether or not a drug is dispensed according to a prescription using a photographed image of the drug for the purpose of preventing misjudgment of drugs with similar shapes, colors, and sizes. is described. Specifically, in the technique described in Patent Document 1, a drug image (registered image) of a drug to be dispensed according to a prescription and a drug similar to the drug is acquired from a drug database, and the acquired drug image and the drug to be dispensed are acquired. This is compared with the photographed image of the drug taken.

JP 2014-64836 A

Here, with the technique described in Patent Literature 1, it is necessary to collate all combinations of each drug packaged in a sachet and each drug image acquired from a database. For example, in the technique described in Patent Document 1, when two drugs, drug A and drug B, are dispensed in one sachet, and there is a drug α similar to A, Three drug images (drug images A, B, and α) are acquired, and a total of 12 (2×3) times of matching between drug A and drug B and the three drug images is required, and efficient matching is required. Not done.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide a drug identification device capable of performing accurate and efficient identification even when drugs having similar drugs are identified. It is to provide an identification method and program.

A drug identification device that is one aspect of the present invention for achieving the above object is a drug identification device that identifies a first drug when the first drug has a similar second drug, A first registered image, which is an image of one drug, and a second registered image, which is an image of a second drug if a second drug is present, are stored in association with the first registered image. A drug database, an image input unit for inputting a target image of a drug to be identified, a drug region extraction unit for extracting a drug region in which the drug is shown from the target image, and a drug database corresponding to the first drug a registered image acquisition unit that acquires the first registered image and the second registered image; a second recognition unit that performs similarity recognition with the second registered image for the drug region recognized as being similar to the first registered image in the first recognition unit; and a drug identification unit that identifies whether the drug is the first drug based on the recognition result of the recognition unit.

According to this aspect, identification is performed based on the result of similarity recognition between the first registered image and the second registered image, and the similarity recognition with the second registered image is performed with the second registered image. Accurate and efficient identification can be performed because it is performed for drug regions that are considered to be similar.

Preferably, the first recognition unit calculates a first degree of similarity between the drug region and the first registered image, and performs similarity recognition by comparing the first degree of similarity with a threshold value.

Preferably, the second recognition unit calculates a second degree of similarity between the drug region and the second registered image.

Preferably, when the first degree of similarity is greater than the second degree of similarity, the drug identification unit identifies the drug appearing in the drug region as the first drug, and the first degree of similarity is the second degree of similarity. If the degree of similarity is less than or equal to 2, then the drug shown in the drug region is identified as not the first drug.

Preferably, a first drug has a plurality of similar second drugs.

Preferably, the second recognition unit calculates a second degree of similarity between the drug region and each of the plurality of second registered images.

Preferably, when the first degree of similarity is greater than each of the plurality of second degrees of similarity, the drug identification unit identifies the drug appearing in the drug region as the first drug, and identifies the drug as the first drug. If the degree of similarity is equal to or less than at least one of a plurality of second degrees of similarity, the drug shown in the drug region is identified as not the first drug.

Preferably, the similarity recognition of the first recognition part and the second recognition part is performed by template matching.

A drug identification method according to another aspect of the present invention is a drug identification method for identifying a first drug when the first drug has a similar second drug, and is an image of the first drug. From a drug database in which a first registered image and a second registered image, which is an image of a second drug if present, are stored in association with the first registered image, a first registered image is obtained. a registration image acquisition step of acquiring a first registration image and a second registration image corresponding to the drug; an image input step of inputting a target image of the drug to be identified; a drug region extracting step of extracting a drug region in the drug region; a first recognition step of performing similarity recognition between the drug region extracted in the drug region extracting step and a first registered image corresponding to the drug region; A second recognition step of performing similarity recognition with the second registered image for the drug region recognized as similar to the first registered image, and based on the recognition results of the first recognition step and the second recognition step , and a drug identification step of identifying whether the drug is the first drug.

A program that is another aspect of the present invention is a program that causes a computer to execute a drug identification step for identifying a first drug when the first drug has a similar second drug, wherein the first drug from a drug database in which a first registered image that is an image of and a second registered image that is an image of a second drug, if present, is stored in association with the first registered image , a registered image acquisition step of acquiring a first registered image and a second registered image corresponding to the first drug; an image input step of inputting a target image of the drug to be identified; a drug region extraction step of extracting a drug region in which a drug is shown; a first recognition step of performing similarity recognition between the drug region extracted in the drug region extraction step and a first registered image corresponding to the drug region; a second recognition step of performing similarity recognition with the second registered image for the drug region recognized as similar to the first registered image in the recognition step; and recognition of the first recognition step and the second recognition step and a drug identification step of identifying whether the drug is the first drug based on the result.

According to the present invention, identification is performed based on the result of similarity recognition between the first registered image and the second registered image, and the similarity recognition with the second registered image is performed with the second registered image. Accurate and efficient identification can be performed because it is performed for drug regions that are considered to be similar.

FIG. 1 is a diagram showing a similarity distribution. FIG. 2 is a diagram showing a similarity distribution. FIG. 3 is a diagram showing the configuration of the package inspection support system. FIG. 4 is a diagram showing how an image of a sachet is acquired under illumination. FIG. 5 is another diagram showing how an image of a sachet is acquired under illumination. FIG. 6 is a block diagram showing a configuration example of a drug identification device. FIG. 7 is a flow chart describing the drug identification method. FIG. 8 is a diagram showing a storage configuration example of a drug database. FIG. 9 is a diagram showing a target image. FIG. 10 is a flowchart for explaining similarity calculation. FIG. 11 is a flow chart showing similarity recognition by the first recognition unit, similarity recognition by the second recognition unit, and drug identification. FIG. 12 is a diagram showing a target image. FIG. 13 is a flow chart showing similarity recognition by the first recognition unit, similarity recognition by the second recognition unit, and drug identification.

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of a drug identification device, a drug identification method, and a program according to the present invention will be described below with reference to the accompanying drawings.

<Similar drugs>
First, it will be described that accurate identification of the drug α becomes difficult when the drug α has a similar drug β. In the following example, a case will be described in which the drug α is placed in the sachet according to the prescription.

Drug α has drug β, which is a similar drug similar in size, color, and imprint (printing). At this time, the drug α should be prescribed, but even if the drug β is erroneously contained in the sachet, the drug β may be mistakenly identified as the drug α.

1 and 2 are diagrams showing similarity distributions between an image of a drug to be identified and a registered image (master image). The vertical axis of FIGS. 1 and 2 indicates the application frequency, and the horizontal axis indicates the degree of similarity with the registered image. Note that the registered image is an image of a drug prepared in advance for collating the drug. The registered image is stored in the drug database, and the drug can be specified by comparing the registered image with the photographed drug image or the enhanced drug image.

FIG. 1 is a diagram showing similarity distributions of a drug α and a drug γ that is not similar to the drug α. A similarity distribution 53 between the image (drug region) of the drug α and the registered image α of the drug α is shown, and the similarity between the image (drug region) of the drug γ and the registered image α of the drug α A distribution 51 is shown.

There is a certain range in the calculated similarity value between the registered image α and the image of the drug α, but the calculated similarity is relatively high (see the similarity distribution 53). On the other hand, there is a certain range in the calculated similarity value between the registered image α and the image of the drug γ, but the calculated similarity is relatively low (see the similarity distribution 51). However, since the drug α and the drug γ are not similar, the similarity distribution 51 and the similarity distribution 53 do not overlap.

Therefore, by setting the threshold value Th1 for identifying the drug α by similarity between the similarity distributions 51 and 53 as shown in FIG. 1, the drug α can be accurately identified from the registered image α. be able to.

On the other hand, in distinguishing between the drug α and the drug β, which is a similar drug, it is difficult to set the threshold as described above, as described below.

FIG. 2 is a diagram showing similarity distributions between the registered image α and the drug α, and between the registered image α and the drug β. A similarity distribution 57 shows a similarity distribution between an image (drug region) photographed of the drug α and the registered image α, and a similarity distribution 55 shows a similarity distribution between an image (drug region) photographed of the drug β and the registered image α. Similarity distribution is shown. There is a certain range in the calculated similarity value between the registered image α and the image of the drug α, but the calculated similarity is relatively high (see the similarity distribution 57). On the other hand, there is a certain range in the calculated similarity value between the registered image α and the image of the drug β, but the calculated similarity is relatively low (see the similarity distribution 55). Also, since the drug α and the drug β are similar in size, color, and imprint (printing), the similarity distribution 55 and the similarity distribution 57 partially overlap.

Since the similarity distribution 55 and the similarity distribution 57 partially overlap in this manner, the threshold Th1 cannot be set between the two similarity distributions as shown in FIG. On the other hand, if a similarity threshold Th2 (see FIG. 2) for identifying drug α is set, drug α can be identified as drug α, but drug β may also be erroneously identified as drug α. end up Further, when the similarity threshold Th3 (see FIG. 2) is set, the drug A may not be identified as the drug A in some cases.

Thus, it is difficult to determine a similarity threshold value for accurately discriminating a drug α having a similar drug β whose similarity distributions partially overlap.

Therefore, by using the present invention, it is possible to perform accurate and efficient identification even when identifying drugs having similar drugs.

<Drug identification device>
FIG. 3 is a diagram showing the configuration of a package inspection support system 1 equipped with the drug identification device 2 (FIG. 6) of the present invention. The package inspection support system 1 includes a processing unit 10, a storage unit (drug database 20A (see FIG. 6)) 20, a display unit 30, an operation unit 40, a transport mechanism 50, an illumination unit 12, and a camera 15A. , a camera 15B, and a prescription reader 16. In addition, the unpackaged inspection support system 1 assists inspection as to whether or not the medicine is properly packaged according to the prescription. Specifically, the drug identification device 2 identifies the packaged drugs and displays the identification result on the display unit 30 (monitor 32), thereby assisting the user's audit. The example of using the drug identification device 2 is not limited to the single packaging inspection. For example, when a patient brings a medicine that has been prescribed at another hospital and brought with him/her at the time of hospitalization, the medicine identification device 2 may be used to identify the medicine that the patient has brought. Moreover, the drugs identified by the drug identification device 2 are not limited to those prescribed by a prescription. For example, even if the drug identification device 2 does not have a prescription, if there is information (information that specifies the first drug) that can identify an identification target (first drug), the drug identification device 2 can identify the drug. can be identified.

The cameras 15A and 15B are composed of digital cameras. As shown in FIG. 4, the camera 15A is arranged above (the +Z side in FIG. 4) a medicine bandage PB in which the sachets TP (medicine packages) are continuously formed. (-Z side) of the camera 15B, and photographs the drug packed in one sachet TP from above and below (in a plurality of different directions). The sachet TP (medicine bandage PB) is conveyed in the +X direction of FIG. 4 (the axis along the longitudinal direction of the medicine bandage PB; the direction of the arrow in FIG. A plurality of light sources 13 provided illuminate the sachet TP from four directions. In FIG. 5, the distances (d1, d2, d3, d4) between each of the plurality of light sources 13 and the imaging optical axes PA of the cameras 15A, 15B are the same. That is, the plurality of light sources 13 and the photographing optical axis PA are equidistant (d1=d2=d3=d4).

A prescription reader 16 reads prescription information. For example, by OCR (Optical Character Recognition), information such as the patient's name, the prescribed medicine and its quantity is read from the prescription written on paper. When a barcode or the like indicating information about the prescribed medicine is recorded on the prescription, information such as the prescribed medicine and its quantity may be read from the barcode. Alternatively, the user may read the prescription and input prescription information using an input device such as a keyboard of the operation unit 40 .

FIG. 6 is a block diagram showing a configuration example of the drug identification device 2. As shown in FIG. The drug identification device 2 is mainly composed of a processing section 10 and a storage section 20 .

The processing unit 10 includes an image input unit 10A, a drug region extraction unit 10B, a registered image acquisition unit 10C, a first recognition unit 10D, a second recognition unit 10E, and a drug identification unit 10F. These functions refer to data stored in EEPROM (Electronically Erasable and Programmable Read Only Memory: non-temporary recording medium) by devices such as the CPU (Central Processing Unit) and various electronic circuits installed in the computer. is realized. These functions are performed by executing a drug identification program stored in a ROM (Read Only Memory: non-temporary recording medium) of the storage unit 20 or the like. During processing, a RAM (Random Access Memory) or the like is used as a temporary storage area and work area. Note that illustration of these devices is omitted in FIG.

The processing unit 10 is also connected to the storage unit 20, and can freely retrieve and rewrite the programs stored in the storage unit 20 and the data of the drug database 20A.

<Drug identification method>
FIG. 7 is a flow chart for explaining a drug identification method (drug identification step) using the drug identification device 2 . A detailed description of each step will be given later.

First, the prescription is read by the prescription reader 16 to obtain prescription information. Then, based on the prescription information, the registered image acquisition unit 10C acquires the first registered image and the second registered image of the medicine that can be dispensed according to the prescription (registered image acquisition step: step S10). Here, the first registered image is the registered image of the first drug dispensed according to the prescription information. Also, the second registered image is a registered image of a second drug similar to the first drug.

Next, a target image PI (see FIG. 9) obtained by photographing one sachet TP is input to the image input unit 10A (image input step: step S11). Thereafter, the drug region extraction unit 10B extracts the drug region R1 and the drug region R2 in the target image PI (see FIG. 9) (drug region extraction step: step S12). Then, the first recognition unit 10D recognizes the drug region R1 and the drug region R2 as being similar to the first registered image (first recognition step: step S13). Then, the first recognition unit 10D performs similarity recognition between the drug region R1 and the drug region R2 and the first registered image (step S14). When the first recognition unit 10D recognizes that the image is not similar to the first registered image, the drug identification unit 10F identifies the drug based on the recognition result of the first recognition unit 10D. On the other hand, when the first recognition unit 10D recognizes that the drug region is similar to the first registered image, the second recognition unit 10E recognizes the drug region that is recognized as similar to the first registered image. (second recognition step: step S15). After that, the drug identification unit 10F identifies the drug based on the recognition results of the first recognition unit 10D and the second recognition unit 10E (drug identification step: step S16).

Next, each step of the drug identification method using the package inspection support system 1 equipped with the drug identification device 2 described above will be described in detail.

<<Registered Image Acquisition Step>>
The registered image obtaining step (step S10) is performed by the registered image obtaining section 10C. The registered image acquisition unit 10C acquires a first registered image and a second registered image corresponding to the first drug based on the prescription information from the drug database 20A. Specifically, the registered image acquiring unit 10C acquires the prescription information acquired by the prescription reader 16, and obtains a first registered image of the first drug dispensed according to the prescription information and a first registered image similar to the first drug. A second registered image of the second drug is obtained from the drug database 20A. In the above description, the registered image acquiring unit 10C acquires the first registered image and the second registered image corresponding to the first drug based on the prescription information by the prescription reader 16. However, the manner in which the registered image acquisition unit 10C acquires the first registered image and the second registered image from the drug database 20A is not limited to the above example. For example, even if the user inputs prescription information using the operation unit 40 and the registered image acquiring unit 10C acquires the first registered image and the second registered image from the drug database 20A based on the entered prescription information. good.

FIG. 8 is a diagram showing a storage configuration example of the drug database 20A.

In the drug database 20A, "medicine name", "registered image" and "similar drug" are stored in association with each other. The name of the drug A is stored in association with the surface image of the drug A and the drug X as the similar drug. The surface image of drug A has an "A" imprint on the surface. Also, a drug X is a drug similar to the drug A. The drug name of drug B is stored in association with the surface image of drug B and the fact that there is no similar drug. The surface image of drug B has a "B" imprinted on the surface. The drug name of drug C is stored in association with the surface image of drug C and drug Y and drug Z as similar drugs. The surface image of drug C has a "C" imprinted on the surface. Drugs similar to drug C are drug Y and drug Z. Thus, in the drug database 20A, drug names, registered images, and similar drugs are stored in association with each other. Note that the storage configuration of the drug database 20A shown in FIG. 8 is an example and is not limited to this. For example, the registered image may have a front side image of the medicine and a back side image of the medicine. In this way, by having the front side image and the back side image in the registered image, it is possible to perform similarity recognition in more detail.

For example, when the drug A and the drug B are to be placed in one sachet TP based on the prescription, the registered image acquiring unit 10C acquires the registered image A (first registered image) of the drug A. and a registered image B (first registered image) of the drug B is acquired. Further, since the drug X is similar to the drug A, the registered image acquiring unit 10C acquires the registered image X of the drug X (second registered image).

<<Image input step and drug region extraction step>>
The image input step (step S11) is performed by the image input unit 10A. A target image of a drug to be identified is input to the image input unit 10A. For example, the target image is an image obtained by photographing one sachet TP containing a drug when identifying that the drug is packaged as a single package according to the prescription. Here, the target image may be composed of a single photographed image of the sachet TP, or may be an image obtained by emphasizing the markings or printing of the drug A and the drug B in one or more photographed images by image processing. There may be.

The drug region extraction step (step S12) is performed by the drug region extraction unit 10B. The drug region extracting unit 10B extracts (cuts out) a drug region in which a drug to be identified is shown from the target image for each drug. The drug region extraction unit 10B extracts the drug region based on, for example, the color difference between the drug and the background.

FIG. 9 is a diagram showing the target image PI input to the image input unit 10A. The target image PI is photographed by the camera 15A, and is subjected to enhancement processing by a marking (printing) enhancement processing section (not shown) of the processing section 10. FIG. The drug region extraction unit 10B extracts the drug region R1 of the drug A and the drug region R2 of the drug B, respectively. In addition, in the single packaging inspection support system 1, the camera 15A and the camera 15B can photograph the sachet TP from both sides to obtain images of the front and back sides of the medicine without omission. However, in the example shown in FIG. 9, the images were taken by the camera 15A, and the images of the surfaces of the medicine A and the medicine B were obtained.

<<first recognition step, second recognition step and drug identification step>>
The first recognition step (step S13) is performed by the first recognition unit 10D. The first recognition unit 10D performs similarity recognition between the drug region extracted by the drug region extraction unit 10B and the corresponding first registered image. Specifically, the first recognition unit 10D calculates a first degree of similarity between the drug region and the first registered image, and performs similarity recognition by comparing the first degree of similarity with a threshold value.

The second recognition step (step S15) is performed by the second recognition section 10E. The second recognition unit 10E performs similarity recognition with the second registered image on the drug region recognized as being similar to the first registered image in the first recognition unit 10D. Specifically, the second recognition unit 10E calculates a second degree of similarity between the drug region and the second registered image.

The drug identification step (step S16) is performed by the drug identification unit 10F. The drug identification unit 10F identifies whether or not the drug is the first drug based on the recognition results of the first recognition unit 10D and the second recognition unit 10E. Specifically, when the first degree of similarity is greater than the second degree of similarity, the drug identification unit 10F identifies the drug shown in the drug region as the first drug, and the first similarity is less than or equal to the second degree of similarity, the drug shown in the drug region is identified as not the first drug.

FIG. 10 is a flowchart for explaining similarity calculation performed by the first recognition unit 10D. Although the second recognition unit 10E also calculates the similarity, the similarity is calculated in the same manner as the first recognition unit 10D, so the explanation is omitted.

First, the first recognition unit 10D recognizes whether or not the medicine region and the first registered image are different in size or shape (circumscribing rectangle) (step S20). Here, if either the size or the shape (circumscribing rectangle) of the drug region and the first registered image are different, the similarity is calculated as "0" (step S24). Note that the first recognition unit 10D recognizes whether or not the size or shape (circumscribing rectangle) differs based on the difference in feature amount between the drug region and the first registered image. For example, the first recognition unit 10D recognizes that the drug region and the first registered image are different from each other when the difference in feature amount between the drug region and the first registered image is equal to or greater than a threshold.

Next, the first recognition unit 10D recognizes whether or not the medicine region and the first registered image have different brightness (brightness) (step S21). Here, if the brightness (brightness) of the drug region and the first registered image is different, the first recognition unit 10D calculates the degree of similarity as "0" (step S24). Note that the first recognition unit 10D recognizes whether or not the brightness is different based on the difference in brightness between the drug region and the first registered image. For example, when the difference in brightness between the drug region and the first registered image is equal to or greater than a threshold, the first recognition unit 10D recognizes that the two are different.

Next, the first recognition unit 10D recognizes whether or not the medicine region and the first registered image have different colors (step S22). Here, the first recognition unit 10D recognizes that the degree of similarity is "0" when the medicine region and the first registered image have different colors (step S24). Note that the first recognition unit 10D recognizes whether or not the colors are different based on the difference in R (red), G (green), and B (blue) values between the drug region and the first registered image. do. For example, when the color difference between the drug region and the first registered image is equal to or greater than a threshold, the first recognition unit 10D recognizes that the two are different.

Next, the first recognition unit 10D calculates the degree of similarity between the drug region and the first registered image (step S22). The first recognition unit 10D calculates similarity by template matching. For example, the degree of similarity is 1.0 when it is most similar (when it is the same), and is calculated lower than 1.0 depending on the degree of difference.

Thus, the first recognition unit 10D calculates the first degree of similarity between the drug region and the first registered image. The second recognition unit 10E may calculate the degree of similarity in the same manner as the first recognition unit 10D. , the similarity may be directly calculated by template matching from the beginning (step S23 in FIG. 10).

Next, with respect to similarity recognition by the first recognition unit 10D, similarity recognition by the second recognition unit 10E, and drug identification by the drug identification unit 10F, identification of each of the drug A and the drug B will be specifically described.

FIG. 11 is a flowchart showing similarity recognition by the first recognition unit 10D, similarity recognition by the second recognition unit 10E, and drug identification.

First, identification of the drug region R1 in the target image PI will be described.

The first recognition unit 10D performs similarity recognition between the drug region R1 and the registered image A and the registered image B (step S30). Specifically, since the drug region R1 and the registered image A have the same size (shape), luminance (brightness), and color, the first recognition unit 10D calculates the degree of similarity by template matching ( (see step S23 in FIG. 10). On the other hand, the first recognition unit 10D calculates the degree of similarity as zero because the drug region R1 and the registered image B have different sizes.

Then, the first recognition unit 10D performs similarity recognition by comparing the calculated degree of similarity with the threshold value ThA. Since the degree of similarity between the drug region R1 and the registered image A is greater than or equal to the threshold ThA, it is recognized that the drug region R1 and the registered image A are similar. It should be noted that the threshold ThA is preferably set lower than the point of extracting the medicine A as thoroughly as possible. For example, the threshold ThA is set like the threshold Th2 described in FIG. On the other hand, the first recognition unit 10D recognizes that the drug region R1 and the registered image B are not similar because the degree of similarity between the drug region R1 and the registered image B is zero and is smaller than the threshold value ThA. It is identified as not drug B (step S35).

After that, the second recognition unit 10E determines whether or not there is a drug similar to the drug A (step S31), and if there is a drug similar to the drug A, calculates a second degree of similarity between the drug region R1 and the registered image X. (Step S32). Then, the drug identification unit 10F identifies whether the drug region R1 is the drug A based on the first degree of similarity and the second degree of similarity (step S33). Specifically, in drug identification, when the first degree of similarity is greater than the second degree of similarity, the drug region R1 is identified as the drug A (step S34). On the other hand, if the first degree of similarity is less than or equal to the second degree of similarity, it is determined that the drug region R1 is not the drug A (step S34).

Next, the identification of the drug region R2 in the target image PI will be similarly described with reference to FIG.

The first recognition unit 10D performs similarity recognition between the drug region R2 and the registered image A and the registered image B (step S30). Specifically, since the medicine region R2 and the registered image B have the same size (shape), color, and stamp (printing), the first recognition unit 10D calculates the degree of similarity by template matching (Fig. 10 , see step S23). On the other hand, the first recognition unit 10D calculates the degree of similarity as zero because the drug region R2 and the registered image A have different sizes.

Then, the first recognition unit 10D performs similarity recognition by comparing the calculated degree of similarity with the threshold value ThA. Since the degree of similarity between the drug region R2 and the registered image B is greater than or equal to the threshold ThA, it is recognized that the drug region R2 and the registered image B are similar. Then, the second recognition unit 10E determines whether there is a drug similar to the drug B (step S31), and since there is no drug similar to the drug B, the second recognition unit 10E calculates the second similarity. do not. After that, the drug identification unit 10F identifies the drug shown in the drug region R2 as the drug B based on the similarity recognition of the first recognition unit 10D (step S34). On the other hand, the first recognition unit 10D recognizes that the drug region R2 and the registered image A are not similar because the degree of similarity between the drug region R2 and the registered image A is zero and is smaller than the threshold value ThA. It is identified as not drug A (step S35).

As described above, in the present invention, first, the first recognition unit 10D recognizes the similarity between the drug region and the corresponding first registered image. Then, in this aspect, similarity recognition with the second registered image is performed in the first recognition unit 10D for the drug region recognized as being similar to the first registered image. Accordingly, in this aspect, since the similarity recognition with the second registered image is performed on the drug region considered to be similar to the second registered image, accurate and efficient identification can be performed.

<Other examples>
<<If there are multiple similar drugs>>
In the above description, the case where drug A (first image) has one similar drug (medicine X) has been described. However, the present invention is not limited to this example, and is also applicable when there are multiple similar drugs.

FIG. 12 is a diagram showing the target image PJ input to the image input unit 10A. The target image PJ is obtained in the same manner as the target image PI described with reference to FIG. Then, the drug region extraction unit 10B extracts the drug region R3 of the drug C. FIG.

FIG. 13 is a flowchart showing similarity recognition by the first recognition unit 10D, similarity recognition by the second recognition unit 10E, and drug identification by the drug identification unit 10F. Identification of the drug C will be described along the flowchart of FIG. 11 . In addition, according to the prescription information, the drug C, which is the first drug, is to be put in one sachet TP. Image C, registered image Y, and registered image Z are acquired from drug database 20A.

The first recognition unit 10D performs similarity recognition between the drug region R3 and the registered image C (step S40). Specifically, since the drug region R3 and the registered image C have the same size (shape), brightness (brightness), and color, the first recognition unit 10D calculates the degree of similarity by template matching ( (see step S23 in FIG. 10).

After that, the first recognition unit 10D performs similarity recognition by comparing the calculated degree of similarity with the threshold value ThA. Since the degree of similarity between the drug region R3 and the registered image C is greater than or equal to the threshold ThA, it is recognized that the drug region R3 and the registered image C are similar.

After that, the second recognition unit 10E calculates a second degree of similarity (1) between the drug region R3 and the registered image Y (step S41). The second recognition unit 10E also calculates a second degree of similarity (2) between the drug region R3 and the registered image Z (step S42). In this way, the second recognition unit 10E calculates the degree of similarity for each of the registered images of similar drugs. Then, the drug identification unit 10F performs identification based on each of the first degree of similarity and the second degree of similarity. Specifically, the drug identification unit 10F determines whether or not the first degree of similarity is greater than the second degree of similarity (1) (step S43). The drug identification unit 10F identifies that the drug region R3 is not the drug C when the first degree of similarity is not greater than the second degree of similarity (1). On the other hand, if the first degree of similarity is greater than the second degree of similarity (1), then the drug identification unit 10F determines whether the first degree of similarity is greater than the second degree of similarity (2). It is determined whether or not (step S44). If the first degree of similarity is not greater than the second degree of similarity (2), the drug identification unit 10F identifies that the drug region R3 is not the drug C (step S46). On the other hand, when the first degree of similarity is greater than the second degree of similarity (2), drug region R3 is identified as drug C (step S45).

As described above, even when a drug having two or more similar drugs is identified, it is possible to accurately identify the drug based on the recognition results of the first recognition unit 10D and the second recognition unit 10E. Efficient drug identification can be performed.

Although examples of the present invention have been described above, it goes without saying that the present invention is not limited to the above-described embodiments, and that various modifications are possible without departing from the spirit of the present invention.

1: Package inspection support system 2: Drug identification device 10: Processing unit 10A: Image input unit 10B: Drug area extraction unit 10C: Registered image acquisition unit 10D: First recognition unit 10E: Second recognition unit 10F: Drug identification unit 12: illumination unit 13: light source 15A: camera 15B: camera 16: prescription reader 20: storage unit 20A: drug database 30: display unit 40: operation unit 50: transport mechanism

Claims (10)

A drug identification device for identifying a first drug when the first drug has a similar second drug,
A first registered image that is an image of the first drug and a second registered image that is an image of the second drug when the second drug is present are associated with the first registered image. a drug database stored in
an image input unit for inputting a target image of the drug to be identified;
a drug region extraction unit that extracts a drug region in which the drug is shown from the target image;
a registered image acquiring unit that acquires the first registered image and the second registered image corresponding to the first drug from the drug database;
a first recognition unit that performs similarity recognition between the drug region extracted by the drug region extraction unit and the corresponding first registered image;
a second recognition unit that performs similarity recognition with the second registered image for the drug region recognized as being similar to the first registered image in the first recognition unit;
a drug identification unit that identifies whether the drug is the first drug based on the recognition results of the first recognition unit and the second recognition unit;
with
The first recognition unit calculates a first similarity between the drug region and the first registered image, and sets the first similarity to a value for identifying the first drug. a drug identification device that performs the similarity recognition by comparing with the threshold value . The drug identification device according to claim 1 , wherein the second recognition unit calculates a second degree of similarity between the drug region and the second registered image. The drug identification unit identifies the drug appearing in the drug region as the first drug when the first degree of similarity is greater than the second degree of similarity, and 3. The medicine identification device according to claim 2 , wherein the medicine reflected in the medicine area is identified as not the first medicine when the degree of similarity is equal to or less than the second degree of similarity. 2. The medicine identification device according to claim 1, wherein the first medicine has a plurality of similar second medicines. The drug identification device according to claim 4 , wherein the second recognition unit calculates a second degree of similarity between the drug region and each of the plurality of second registered images. The drug identification unit identifies the drug reflected in the drug region as the first drug when the first degree of similarity is greater than each of the plurality of second degrees of similarity, and 6. The medicine according to claim 5 , wherein the medicine appearing in the medicine area is identified as not the first medicine when the first degree of similarity is equal to or less than at least one of the plurality of second degrees of similarity. Drug identification device. The drug identification device according to any one of claims 1 to 6 , wherein the similarity recognition by the first recognition section and the second recognition section is performed by template matching. A drug identification method for identifying a first drug when the first drug has a similar second drug,
A first registered image that is an image of the first drug and a second registered image that is an image of the second drug when the second drug is present are associated with the first registered image. a registered image acquiring step of acquiring the first registered image and the second registered image corresponding to the first drug from a drug database stored in the
an image input step of inputting a target image of the drug to be identified;
a drug region extracting step of extracting a drug region in which the drug is shown from the target image;
a first recognition step of performing similarity recognition between the drug region extracted in the drug region extraction step and the corresponding first registered image;
a second recognition step of performing similarity recognition with the second registered image for the drug region recognized as being similar to the first registered image in the first recognition step;
a drug identification step of identifying whether the drug is the first drug based on the recognition results of the first recognition step and the second recognition step;
including
In the first recognition step, a first degree of similarity between the drug region and the first registered image is calculated, and a value for identifying the first drug is set to the first degree of similarity. A drug identification method for performing the similarity recognition by comparing with the threshold value . A program for causing a computer to execute a drug identification step for identifying the first drug when the first drug has a similar second drug,
A first registered image that is an image of the first drug and a second registered image that is an image of the second drug when the second drug is present are associated with the first registered image. a registered image acquiring step of acquiring the first registered image and the second registered image corresponding to the first drug from a drug database stored in the
an image input step of inputting a target image of the drug to be identified;
a drug region extracting step of extracting a drug region in which the drug is shown from the target image;
a first recognition step of performing similarity recognition between the drug region extracted in the drug region extraction step and the corresponding first registered image;
a second recognition step of performing similarity recognition with the second registered image for the drug region recognized as being similar to the first registered image in the first recognition step;
a drug identification step of identifying whether the drug is the first drug based on the recognition results of the first recognition step and the second recognition step;
including
In the first recognition step, a first degree of similarity between the drug region and the first registered image is calculated, and a value for identifying the first drug is set to the first degree of similarity. A program for causing a computer to execute the drug identification step of performing the similarity recognition by comparing with the determined threshold value . A non-transitory computer-readable recording medium that causes a computer to execute the program according to claim 9 when instructions stored in the recording medium are read by the computer.

Images