JP2020184145A - Article recognition system - Google Patents

Abstract

To implement a technique which allows appropriate recognition of identification information attached to an outer surface of each of articles constituting an article group placed on a support.SOLUTION: An article recognition system 1 comprises: a first imaging apparatus 21 and a second imaging apparatus 22 being two types of imaging apparatuses 20 which take an article group WG being a group of a plurality of articles W placed on a support 4, as an imaging object to capture images of an outer surface of the article group WG; an image division unit which divides the image of the article group WG captured by the second imaging apparatus 22 into unit images being images of individual articles W constituting the article group WG on the basis of the image of the article group WG captured by the first imaging apparatus 21; and an identification information recognition unit which recognizes identification information attached to outer surfaces of the articles W constituting the article group WG on the basis of the unit images.SELECTED DRAWING: Figure 1

Description

The present invention relates to an article recognition system that recognizes identification information attached to the outer surface based on a captured image of the outer surface of the article.

An example of the article recognition system as described above is disclosed in Japanese Patent Application Laid-Open No. 2015-41371 (Patent Document 1). Patent Document 1 describes that the serial number and the expiration date attached to the outer surface are recognized based on the image captured on the outer surface of the article by the imaging device.

Japanese Unexamined Patent Publication No. 2015-41371

By the way, in a physical distribution facility, an article may be handled in a support unit (for example, a pallet unit) on which an article is placed. In such a case, it is conceivable that each of the articles constituting the article group mounted on the support is to be recognized as the identification information, but Patent Document 1 does not describe this point.

Therefore, it is desired to realize a technique capable of appropriately recognizing the identification information attached to the outer surface of each of the articles constituting the article group mounted on the support.

The article recognition system according to the present disclosure is an article recognition system that recognizes identification information attached to the outer surface based on an image captured on the outer surface of the article, and is a group of a plurality of articles mounted on a support. The second image based on the first image pickup device and the second image pickup device, which are two types of image pickup devices that image the outer surface of the article group, and the image captured by the first image pickup device. An image dividing unit that divides an image captured by the image pickup device into a unit image that is an image of each article that constitutes the article group, and an outer surface of the article that constitutes the article group based on the unit image. It is provided with an identification information recognition unit that recognizes the identification information.

According to this configuration, when each of the articles constituting the article group mounted on the support is to be recognized as the identification information, the identification information can be recognized based on the unit image which is an image of each article. Therefore, the identification information attached to the outer surface of each of the articles constituting the article group can be appropriately recognized.

Then, according to the above configuration, the division of the captured image of the article group into unit images by the imaging device (specifically, the second imaging device) is performed by an imaging device of a type different from that of the imaging device (specifically). Can be performed based on the captured image of the article group by the first imaging device). Unlike such a configuration, when the image captured by the image pickup device is divided into unit images based on the image captured by the image pickup device, the acquisition condition of the captured image (characteristics of the image pickup device). , Imaging conditions, etc.) must be set in consideration of both recognition of identification information and division into unit images, but acquisition conditions of captured images suitable for recognition of identification information and division into unit images It does not always match the acquisition conditions of the captured image suitable for. On the other hand, according to the above configuration, even when these two acquisition conditions are different, the acquisition condition of the image captured by the first imaging device is set as the acquisition condition suitable for division into unit images. Since the acquisition condition of the captured image by the second imaging device can be set as the acquisition condition suitable for recognizing the identification information, the identification information attached to the outer surface of each of the articles constituting the article group mounted on the support. Is easy to recognize properly.

Further features and advantages of the article recognition system will be clarified from the following description of embodiments described with reference to the drawings.

Perspective view showing a part of the logistics equipment to which the goods recognition system is applied. Perspective view of a group of articles mounted on a support Block diagram showing the schematic configuration of the article recognition system Flowchart showing the processing procedure of the article recognition process Flowchart showing the processing procedure of the identification information recognition process Explanatory drawing of image division processing Explanatory drawing of narrowing down of matching data included in the second category Explanatory drawing of narrowing down of matching data included in the third category The figure which shows the state which the identification information of three categories was recognized

An embodiment of the article recognition system will be described with reference to the drawings.

The article recognition system 1 is a system that recognizes the identification information 10 (see FIG. 2) attached to the outer surface based on the captured image of the outer surface of the article W. In the present embodiment, the article recognition system 1 is applied to the distribution facility 100 as illustrated in FIG. Note that FIG. 1 shows only the vicinity of the receiving point P in the distribution facility 100. The receiving location P is a location where the article W is carried in from the outside of the distribution facility 100. A transport device 40 for transporting the article W is provided at the receiving location P. In the present embodiment, the transport device 40 is configured to transport the support 4 such as a pallet, and the article W is transported by the transport device 40 while being mounted on the support 4. The article W carried out from the receiving location P by the transport device 40 is stored in, for example, an automated warehouse. In the example shown in FIG. 1, at the receiving location P, the support 4 on which the article W is placed is carried into the transport device 40 by the forklift 9.

As shown in FIG. 1, at the imaging position Q set on the transport path of the article W by the transport device 40, the outer surface of the article W is imaged in order to recognize the identification information 10. That is, the image pickup device 20 included in the article recognition system 1 images the outer surface of the article W at the imaging position Q. The image of the outer surface of the article W by the imaging device 20 may be performed while the article W is stopped with respect to the imaging device 20, or may be performed while the article W is moving with respect to the imaging device 20. ..

In the present embodiment, the image pickup apparatus 20 is provided in a fixed state. Therefore, due to the operation of the transport device 40, the state in which the article W is moving in the direction along the transport path of the transport device 40 becomes the state in which the article W is moving with respect to the image pickup device 20. Further, in the present embodiment, a rotation mechanism 3 for rotating the article W around the rotation axis R (here, the axis along the vertical direction V) (including turning, the same applies hereinafter) is provided at the imaging position Q. Therefore, even when the article W is rotating around the rotation axis R by the operation of the rotation mechanism 3, the article W is in a state of moving with respect to the image pickup device 20. Although details are omitted, the rotation mechanism 3 rotates the article W mounted on the support 4 around the rotation axis R by rotating the turntable 3a that supports the support 4 from below around the rotation axis R. It is configured as follows.

The article recognition system 1 recognizes the identification information 10 attached to the outer surface of the article W based on the image captured by the image pickup device 20 on the outer surface of the article W. Here, the article recognition system 1 is attached to each outer surface of the article W constituting the article group WG based on a captured image of the outer surface of the article group WG which is a group of a plurality of articles W mounted on the support 4. Recognize the identification information 10. That is, assuming that the article W of the recognition target (recognition target of the identification information 10) is the recognition object 5, each of the articles W constituting the article group WG mounted on the support 4 is the recognition object 5. The recognized identification information 10 of the article W collates, for example, the type and number of articles W carried in from the outside of the distribution facility 100 with the contents of the transportation order (for example, the contents described in the delivery slip) (for example, the contents described in the delivery slip). Used for inspection work). Further, the recognized identification information 10 of the article W is used, for example, in the work of creating a database of the types and the number of articles W carried in from the outside of the distribution facility 100.

In the present embodiment, as shown in FIG. 2, the article W has a rectangular parallelepiped outer shape. The article W is, for example, a corrugated cardboard case. Article W has six outer surfaces. Specifically, the article W has two first side surfaces S1 facing opposite sides, two second side surfaces S2 facing each other, an upper surface S3, and a lower surface. The two first side surfaces S1 are arranged parallel to each other, and the two second side surfaces S2 are arranged parallel to each other. Further, the first side surface S1 and the second side surface S2 intersect each other at right angles. Then, identification information 10 for identifying the article W is attached to the outer surface of the article W.

The identification information 10 is attached to the outer surface of the article W, for example, by printing or pasting. In FIG. 2, the identification information 10 is simplified and represented by a graphic, but the identification information 10 is represented by, for example, characters, figures, photographs, pictures, symbols, or a combination thereof. The characters include stylized characters (design characters) and decorative characters. In the present embodiment, the identification information 10 is represented in a form in which a person such as a worker can recognize the content of the information. In the example shown in FIG. 2, two identification information 10 (specifically, the first identification information 11 and the second identification information 12) are attached to the first side surface S1, and one identification information is attached to the second side surface S2. 10 (specifically, the third identification information 13) is attached. Identification information 10 may be attached to the upper surface S3. In FIG. 2, by representing the first identification information 11, the second identification information 12, and the third identification information 13 in different figures, it is shown that these three identification information 10 represent information different from each other. .. That is, in the example shown in FIG. 2, different identification information 10 is attached to the plurality of outer surfaces of the article W. The same identification information 10 (identification information 10 representing information having the same content) may be attached to a plurality of outer surfaces of the article W.

As shown in FIGS. 1 and 2, the support 4 is arranged so that the support surface that supports the article W from below is along the horizontal plane. Here, the first horizontal direction H1 and the second horizontal direction H2, which are two horizontal directions orthogonal to each other, are defined with reference to the support 4. The first horizontal direction H1 and the second horizontal direction H2 rotate while maintaining a relationship in which the first horizontal direction H1 and the second horizontal direction H2 are orthogonal to each other in accordance with the rotation of the support 4 around the rotation axis R. Defined in. In the present embodiment, the support 4 has a rectangular shape (square shape) having two sides along the first horizontal direction H1 and two sides along the second horizontal direction H2 in a plan view (direction view along the vertical direction V). Is formed in).

As shown in FIG. 2, in the article W, the first side surface S1 is along one of the first horizontal direction H1 and the second horizontal direction H2, and the second side surface S2 is the first horizontal direction H1 and the second horizontal direction H2. It is placed on the support 4 in the direction along the other side of the. Then, in the present embodiment, a plurality of articles W are arranged side by side in the first horizontal direction H1 and the second horizontal direction H2 on one support 4. Further, in the present embodiment, a plurality of articles W are stacked on one support 4 (in other words, they are arranged so as to be stacked in the vertical direction V). In the present embodiment, a plurality of articles W having the same outer shape are placed on one support 4. In the example shown in FIG. 2, in a plurality of articles W mounted on one support 4, the first side surface S1 is along the first horizontal direction H1 and the second side surface S2 is in the second horizontal direction H2. An article W arranged in a direction along the line and an article W in which the first side surface S1 is arranged along the second horizontal direction H2 and the second side surface S2 is arranged along the first horizontal direction H1 are mixed. ..

The article recognition system 1 includes an image pickup device 20 that images the outer surface of the recognition object 5. Here, the article recognition system 1 includes a first image pickup device 21 and a second image pickup device 22, which are two types of image pickup devices 20. Then, each of the first imaging device 21 and the second imaging device 22 images the outer surface of the article group WG with the article group WG mounted on the support 4 as an imaging target. In the present embodiment, the first imaging device 21 and the second imaging device 22 are configured to image the article group WG from the same side. A part of the first image pickup device 21 and a part of the second image pickup device 22 may be composed of common parts, but even in that case, at least the image pickup device and the optical system are the first image pickup. The device 21 and the second image pickup device 22 are provided separately.

In the present embodiment, the first imaging device 21 is configured to image the article group WG from two or more different directions. Here, the first imaging device 21 is configured to image the article group WG from five different directions. Specifically, the first imaging device 21 is configured to image the article group WG from both sides of the first horizontal direction H1, both sides of the second horizontal direction H2, and the upper side of the vertical direction V. Further, in the present embodiment, the second imaging device 22 is configured to image the article group WG from two or more different directions. Here, the second imaging device 22 is configured to image the article group WG from five different directions. Specifically, the second imaging device 22 is configured to image the article group WG from both sides of the first horizontal direction H1, both sides of the second horizontal direction H2, and the upper side of the vertical direction V.

For example, the image pickup device 20 is configured to include a plurality of cameras that image the article group WG from different directions (hereinafter referred to as "first configuration"), so that the article group WG is imaged from two or more different directions. The image pickup device 20 can be configured as such. Further, the article group WG can also be configured such that the article recognition system 1 includes an image pickup direction changing mechanism 2 (hereinafter, referred to as “second configuration”) that changes the imaging direction of the image pickup device 20 relative to the article group WG. The imaging device 20 can be configured to image from two or more different directions. The image pickup direction changing mechanism 2 rotates (including rotation) about at least one of the article group WG and the image pickup device 20 around an axis intersecting (for example, orthogonal to) the image pickup direction (optical axis direction) of the image pickup device 20. Therefore, the imaging direction of the imaging device 20 with respect to the article group WG is configured to be relatively changed. In the present embodiment, by adopting both the first configuration and the second configuration, the first imaging device 21 and the second imaging device 22 are configured to image the article group WG from five different directions. There is.

Specifically, as shown in FIG. 1, in the present embodiment, the first imaging device 21 sets the first camera 21a that images the article group WG from one side of the specific horizontal direction H and the article group WG in the specific horizontal direction. It includes three first cameras 21a, a first camera 21a that images from the other side of the direction H and a first camera 21a that images the article group WG from the upper side of the vertical direction V. That is, the first imaging device 21 is configured to be provided with these three first cameras 21a so that the article group WG can be simultaneously imaged from three different directions. Further, in the present embodiment, the second imaging device 22 has a second camera 22a that images the article group WG from one side of the specific horizontal direction H, and a second camera that images the article group WG from the other side of the specific horizontal direction H. It is provided with three second cameras 22a, a camera 22a and a second camera 22a that images the article group WG from above in the vertical direction V. That is, the second imaging device 22 is configured to be provided with these three second cameras 22a so that the article group WG can be simultaneously imaged from three different directions. Here, the specific horizontal direction H is a specific horizontal direction. The specific horizontal direction H is defined as a direction fixed in the space where the distribution facility 100 is provided, unlike the first horizontal direction H1 and the second horizontal direction H2. As shown in FIG. 1, in the present embodiment, the specific horizontal direction H is a horizontal direction orthogonal to the transport direction of the article W by the transport device 40.

As described above, in the present embodiment, the rotation mechanism 3 is provided at the imaging position Q, and the article recognition system 1 rotates the article group WG around the rotation axis R at the imaging position Q (specifically, (By rotating 90 degrees), the first state in which the article group WG is arranged so that the first horizontal direction H1 is along the specific horizontal direction H, and the article so that the second horizontal direction H2 is along the specific horizontal direction H. It is configured to be able to switch between the second state in which the group WG is arranged and the second state. In the first state, the article group WG can be imaged from both sides of the first horizontal direction H1 and from the upper side of the vertical direction V by the first imaging device 21 and the second imaging device 22. Further, in the second state, the article group WG can be imaged from both sides of the second horizontal direction H2 and from the upper side of the vertical direction V by the first imaging device 21 and the second imaging device 22. That is, the first imaging device 21 and the second imaging device 22 are configured to image the article group WG from five different directions by performing imaging in the first state and imaging in the second state. There is.

As described above, in the present embodiment, the article recognition system 1 includes an imaging direction changing mechanism 2 that relatively changes the imaging direction of the first imaging device 21 and the second imaging device 22 with respect to the article group WG. In the present embodiment, the rotation mechanism 3 described above constitutes the imaging direction changing mechanism 2. That is, the imaging direction changing mechanism 2 is configured to rotate the article group WG around the rotation axis R to relatively change the imaging directions of the first imaging device 21 and the second imaging device 22 with respect to the article group WG. Has been done. Specifically, the imaging directions of the two first cameras 21a and the two second cameras 22a that image the article group WG from both sides in the specific horizontal direction H are orthogonal to the rotation axis R. Therefore, by rotating the article group WG around the rotation axis R, the imaging directions of these four cameras (two first cameras 21a and two second cameras 22a) with respect to the article group WG are changed.

As shown in FIG. 3, the article recognition system 1 includes a control device 30. The control device 30 includes an arithmetic processing unit such as a CPU and peripheral circuits such as a memory, and the control device 30 cooperates with these hardware and a program executed on the hardware such as the arithmetic processing unit. Each function of is realized. The article recognition system 1 includes a storage unit 39, and the control device 30 is configured to be able to extract data from the storage unit 39. Matching data 50 related to the identification information 10 is stored in the storage unit 39. The storage unit 39 includes a storage medium such as a flash memory or a hard disk that can store and rewrite information as a hardware configuration. Although FIG. 3 illustrates a case where the storage unit 39 is configured by a device different from the control device 30, the storage unit 39 may be configured by using the storage device included in the control device 30. Further, the storage unit 39 may be provided in a server or a cloud server capable of communicating with the control device 30.

As shown in FIG. 3, in the present embodiment, the control device 30 includes a captured image acquisition unit 34, an image division unit 31, an identification information recognition unit 32, and an article number recognition unit 33. Although not shown, the control device 30 further includes a functional unit that controls the operation of the transport device 40 and a functional unit that controls the operation of the rotation mechanism 3. These plurality of functional units included in the control device 30 are configured to be able to exchange information with each other. The plurality of functional units included in the control device 30 are at least logically distinguished, and do not necessarily have to be physically distinguished. Further, the plurality of functional units included in the control device 30 need not be realized by common hardware, and may be realized by being divided into a plurality of hardware that can communicate with each other. That is, the control device 30 may be configured by using a plurality of hardware that can communicate with each other instead of one hardware.

Hereinafter, the configuration of each functional unit of the control device 30 will be described, but the technical features of the control device 30 disclosed in the present specification also include an article recognition method for recognizing the identification information 10 attached to the outer surface of the article W. Applicable and such article recognition methods are also disclosed herein.

The captured image acquisition unit 34 is a functional unit that acquires an image captured on the outer surface of the article W (in other words, the recognition object 5) by the image pickup device 20. That is, the captured image acquisition unit 34 acquires image data representing an image of the outer surface of the article W. Here, the captured image acquisition unit 34 acquires the captured image of the article group WG by the imaging device 20. Specifically, the captured image acquisition unit 34 acquires the captured image of the article group WG by the first imaging device 21, and also acquires the captured image of the article group WG by the second imaging device 22.

The image dividing unit 31 is a functional unit that divides the captured image (article group image G) of the article group WG by the imaging device 20 into a unit image GU which is an image for each article W constituting the article group WG. The image division unit 31 acquires the data of the captured image of the article group WG (data of the article group image G) from the captured image acquisition unit 34, and generates the data of the unit image GU. Here, the image dividing unit 31 divides the captured image of the article group WG by the second imaging device 22 into the unit image GU based on the captured image of the article group WG by the first imaging device 21.

As described above, since the image captured by the article group WG by the first imaging device 21 is used for dividing the article group image G into the unit image GU, the acquisition condition of the captured image by the first imaging device 21 (first imaging device 21). (Characteristics, imaging conditions, etc.) are the acquisition conditions suitable for dividing the article group image G into unit images GU. On the other hand, the acquisition conditions of the captured image by the second imaging device 22 (characteristics of the second imaging device 22, the imaging conditions, etc.) are the acquisition conditions suitable for recognizing the identification information 10. The acquisition conditions suitable for dividing into the unit image GU and the acquisition conditions suitable for recognizing the identification information 10 do not always match, but the acquisition conditions for the captured image by the first imaging device 21 and the captured image by the second imaging device 22 By setting the acquisition conditions of the above separately as described above, it becomes easy to appropriately recognize the identification information 10 attached to the outer surface of each of the articles W constituting the article group WG.

In the present embodiment, the image dividing unit 31 captures an image of the article group WG by the second imaging device 22 based on the edge image obtained by performing edge detection processing on the image captured by the article group WG by the first imaging device 21. Is divided into unit image GUs. The edge is detected based on, for example, a change in brightness on the captured image. Since the boundary between the articles W and the boundary between the article W and the background can be recognized based on the edge on the edge image, the boundary between the recognized articles W and the article W can be recognized as schematically shown in FIG. Based on the boundary with the background, the captured image (article group image G) of the article group WG by the second imaging device 22 can be divided into unit image GUs. Since the image dividing unit 31 is configured in this way, in the present embodiment, the acquisition condition of the captured image by the first imaging device 21 is set to the acquisition condition suitable for edge detection. For example, the acquisition condition of using the monochrome imaging device as the first imaging device 21 can be the acquisition condition of the captured image by the first imaging device 21.

The identification information recognition unit 32 is a functional unit that recognizes the identification information 10 attached to the outer surface of the article W (in other words, the outer surface of the recognition object 5). Here, the identification information recognition unit 32 acquires the data of the unit image GU generated by the image division unit 31. Then, the identification information recognition unit 32 recognizes the identification information 10 attached to the outer surface of the article W constituting the article group WG based on the unit image GU. By recognizing the identification information 10 in this way, the identification information recognition unit 32 uses the identification information 10 attached to each outer surface of the articles W constituting the article group WG as a unit image GU which is an image for each article W. Can be properly recognized based on.

The identification information recognition unit 32 performs a matching process between the captured image of the article W (in other words, the recognition target object 5) by the image pickup apparatus 20 and the matching data 50 stored in the storage unit 39, thereby performing the article W (in other words, the article W (in other words)). , Recognizes the identification information 10 attached to the outer surface of the recognition object 5). Here, the identification information recognition unit 32 is the identification information 10 attached to the outer surface of the article W constituting the article group WG by the matching process of the unit image GU and the matching data 50 stored in the storage unit 39. Recognize. Specifically, the identification information recognition unit 32 uses the matching data 50 stored in the storage unit 39 as a matching candidate, and extracts the matching data 50 whose degree of matching with the unit image GU is equal to or higher than the threshold value. Then, the identification information recognition unit 32 recognizes the identification information 10 attached to the outer surface of the article W based on the matching data 50 extracted by the matching process. The matching data 50 is data generated based on the identification information 10, and is stored in the storage unit 39 in association with the information represented by the identification information 10 of the generation source. Therefore, the identification information recognition unit 32 recognizes the information represented by the identification information 10 of the generation source of the matching data 50 associated with the extracted matching data 50 as the identification information 10 attached to the outer surface of the article W. To do. As described above, in the present embodiment, the identification information recognition unit 32 does not recognize the identification information 10 attached to the outer surface of the article W as characters, and the information associated with the extracted matching data 50 is used for the article W. It is configured to be recognized as the identification information 10 attached to the outer surface.

Information (for example, an image) of a plurality of identification information 10 that may be attached to the outer surface of the article W is collected in advance, and matching data 50 is generated and stored based on each of the collected identification information 10. It is stored in part 39. For example, by collecting information (for example, an image) of various articles W in circulation, information on a plurality of identification information 10 that may be attached to the outer surface of the article W is collected. The matching data 50 is, for example, image data representing the identification information 10, or data obtained by performing a feature amount extraction process on the image data representing the identification information 10. In the latter case, in the matching process, the same feature amount extraction process is performed on the image captured by the image pickup apparatus 20 (here, the unit image GU). The acquisition condition of the captured image by the second imaging device 22 is an acquisition condition suitable for the matching process with the matching data 50. Therefore, when the matching data 50 is the data obtained by performing the feature amount extraction process on the image data representing the identification information 10, the acquisition condition of the captured image by the second imaging device 22 is the feature amount of the identification information 10. The acquisition conditions are suitable for the extraction of.

In the present embodiment, in the matching process between the unit image GU and the matching data 50, a partial image extracted from the unit image GU so as to include one identification information 10 is used instead of the entire image of the unit image GU. For example, in the example shown in FIG. 6, the identification information 10 included in the first target image G1 is used by using the first target image G1 which is a partial image extracted from the unit image GU representing the first side surface S1 of the article W. The second target image G2, which is another partial image extracted from the unit image GU representing the first side surface S1 of the article W, is included in the second target image G2. Matching processing is performed to recognize the identification information 10. Further, a matching process for recognizing the identification information 10 included in the third target image G3 by using the third target image G3 which is a partial image extracted from the unit image GU representing the second side surface S2 of the article W. Is done.

As described above, in the present embodiment, the first imaging device 21 is configured to image the article group WG from two or more different directions (specifically, five directions). Therefore, the arrangement state of the articles W constituting the article group WG can be recognized from two or more different viewpoints, and the directions of the articles W constituting the article group WG are not aligned as in the example shown in FIG. 2, for example. Even in the case (that is, when the captured image of the article group WG contains different outer surfaces of the article W), by considering the three-dimensional shape of the article W, the captured image of the article group WG can be transferred to the unit image GU. It is easier to divide properly. Further, in the present embodiment, the second imaging device 22 is configured to image the article group WG from two or more different directions (specifically, five directions). Therefore, the identification information 10 attached to the plurality of outer surfaces of the article W can be recognized, and even when the identification information 10 attached only to a part of the outer surface of the article W exists, the identification information 10 Is easier to recognize. Further, the article W can be appropriately managed by associating the identification information 10 attached to different outer surfaces of the article W with one article group WG or one article W.

The article number recognition unit 33 is a functional unit that recognizes the number of articles W constituting the article group WG based on the captured image of the article group WG by the first imaging device 21. As described above, in the present embodiment, the first imaging device 21 is configured to image the article group WG from two or more different directions (specifically, five directions). Therefore, by recognizing the arrangement state of the articles W constituting the article group WG from two or more different viewpoints, it is easy to recognize the number of articles W constituting the article group WG with high accuracy.

In the present embodiment, the article number recognition unit 33 is based on an edge image (for example, an edge image generated by the image dividing unit 31) obtained by performing edge detection processing on an image captured by the article group WG by the first imaging device 21. The number of articles W constituting the article group WG is recognized. Specifically, the article number recognition unit 33 recognizes and recognizes the loading pattern of the article W by recognizing the boundary between the articles W and the boundary between the article W and the background based on the edge on the edge image. The number of articles W constituting the article group WG is recognized based on the loading pattern. For example, a plurality of assumed stacking patterns are stored in the storage unit 39 in association with the number of articles W constituting each loading pattern, and the recognized stacking pattern of the article W and the stacking stored in the storage unit 39. By performing the matching process with the attachment pattern, the number of articles W constituting the article group WG can be recognized. In addition to the image captured by the first imaging device 21, the article number recognition unit 33 also provides other information (for example, information on the height of the article group WG and information on the weight of the article group WG). Based on this, the configuration may be such that the number of articles W constituting the article group WG is recognized.

Next, the procedure of the article recognition process executed by the article recognition system 1 of the present embodiment will be described with reference to FIG. As shown in FIG. 4, the captured image acquisition unit 34 executes the captured image acquisition process (captured image acquisition step) of acquiring the captured image of the outer surface of the article W (here, the article group WG) by the imaging device 20 (the captured image acquisition step). Step # 01). In the captured image acquisition process, the captured image of the article group WG by the first imaging device 21 and the captured image of the article group WG by the second imaging device 22 are acquired. Further, in the present embodiment, in the captured image acquisition process, captured images of the article group WG are acquired from two or more different directions (here, five directions). In the present embodiment, the article group WG is imaged by the image pickup apparatus 20 while the article group WG is stopped at the imaging position Q.

Next, the image division unit 31 executes an image division process (image division step) for dividing the captured image (article group image G) of the article group WG by the image pickup apparatus 20 into unit image GUs (step # 02). In the image division processing, the image captured by the article group WG by the second imaging device 22 is divided into unit images GU based on the image captured by the article group WG by the first imaging device 21. In the present embodiment, the second imaging device from two or more different directions is used by using the images captured by the first imaging device 21 and the second imaging device 22 imaged from the same side of the article group WG. Image division processing is performed for each of the captured images of the article group WG by 22. For example, using the image captured by the first imaging device 21 of the article group WG from one side of the specific horizontal direction H, the captured image of the article group WG from the one side of the specific horizontal direction H by the second imaging device 22 can be obtained. The unit image is divided into GUs.

Next, the identification information recognition unit 32 executes an identification information recognition process (identification information recognition step) for recognizing the identification information 10 attached to the outer surface of the article W (in other words, the outer surface of the recognition object 5) (step). # 03). In the identification information recognition process, the identification information 10 attached to the outer surface of each of the articles W constituting the article group WG is recognized by performing the matching process for each of the unit image GUs generated by the image division process. In the present embodiment, the identification information recognition process is performed for each of the captured images of the article group WG from two or more different directions. As a result, the identification information 10 attached to the plurality of outer surfaces of the article W can be recognized. Details of the identification information recognition process will be described later.

Next, the article number recognition unit 33 executes an article number recognition process (article number recognition step) for recognizing the number of articles W constituting the article group WG based on the captured image of the article group WG by the image pickup device 20. (Step # 04). In the article number recognition process, the number of articles W constituting the article group WG is recognized based on the image captured by the article group WG by the first imaging device 21. Here, the case where the article number recognition process (step # 04) is performed after the identification information recognition process (step # 03) has been described as an example, but the article number recognition process (step # 04) is performed as the identification information recognition process. It may be performed at the same time as (step # 03) or before that.

Next, the procedure of the identification information recognition process executed by the article recognition system 1 of the present embodiment will be described with reference to FIGS. 5 to 9. Here, as an example, the identification information 10 is classified into three categories T of the first category T1, the second category T2, and the third category T3, and in the identification information recognition process, the first matching process and the second matching process. , And the case where the three matching processes (see FIG. 5) of the third matching process are performed will be described as an example.

As conceptually shown in FIG. 7, the storage unit 39 stores matching data 50 of the identification information 10 belonging to each category T for each of a plurality of categories T in which the identification information 10 is classified. Here, the two categories T included in the plurality of categories T are referred to as the first category T1 and the second category T2. In the example shown in FIG. 7, the plurality of categories T further include a third category T3. In the following, the identification information 10 classified into the first category T1 is referred to as the first identification information 11, the identification information 10 classified into the second category T2 is referred to as the second identification information 12, and the identification information classified into the third category T3 is classified. Information 10 is designated as the third identification information 13 (see FIG. 6).

As shown in FIG. 7, each of the plurality of matching data 50 included in one category T is associated with one or more of the plurality of matching data 50 included in the category T one level lower than the category T. It is stored in the storage unit 39 in this state. Here, the category T one lower than the first category T1 is the second category T2, and the category T one lower than the second category T2 is the third category T3. That is, each of the plurality of matching data 50 included in the first category T1 is stored in the storage unit 39 in a state associated with one or more of the plurality of matching data 50 included in the second category T2. .. Further, each of the plurality of matching data 50 included in the second category T2 is stored in the storage unit 39 in a state of being associated with one or more of the plurality of matching data 50 included in the third category T3. .. In FIG. 7, a line segment (solid line or broken line) indicates that the matching data 50s are associated with each other.

Since the identification information 10 attached to the outer surface of the article W is for identifying the article W, the identification information of a plurality of categories T attached to the outer surface of one article W is generally related to each other. Has sex. Therefore, the association of the matching data 50 as described above is performed based on the association of the identification information 10 among the plurality of categories T.

As shown in FIG. 5, the identification information recognition unit 32 is a matching process for recognizing the first identification information 11 (identification information 10 belonging to the first category T1) attached to the outer surface of the recognition object 5. 1 Matching process (first matching step) is executed (step # 11). Next, the identification information recognition unit 32 is a second matching process (matching process) for recognizing the second identification information 12 (identification information 10 belonging to the second category T2) attached to the outer surface of the recognition object 5. The second matching step) is executed (step # 12). That is, the identification information recognition unit 32 performs the first matching process and then the second matching process. Next, the identification information recognition unit 32 is a third matching process (third matching process) for recognizing the third identification information 13 (identification information 10 belonging to the third category T3) attached to the outer surface of the recognition object 5. The third matching step) is executed (step # 13). That is, the identification information recognition unit 32 performs the third matching process after performing the second matching process. By performing these three matching processes, for example, the three identification information 10 (first identification information 11, second identification information 12, and third identification information 13) in the example shown in FIG. 6 can be recognized.

FIG. 7 is an image (FIG. 7) including the first identification information 11 attached to the outer surface of the recognition object 5 from among the plurality of matching data 50 included in the first category T1 by executing the first matching process. In the example shown in 6, the matching data 50 in which the degree of coincidence with the first target image G1) is equal to or greater than the threshold value is extracted. In the following, the matching data 50 extracted by the matching process will be referred to as extraction matching data 52. In the example shown in FIG. 7, each of the matching data 50 included in the first category T1 is the information "AN" (N = 1, 2) represented by the first identification information 11 of the generation source of each matching data 50. , ...). Then, in the example shown in FIG. 7, since the extraction matching data 52 extracted by the first matching process is associated with the information "A2", the information "A2" is recognized by the execution of the first matching process. It is recognized as the first identification information 11 attached to the outer surface of the object 5.

Then, in the matching process after the second matching process, the matching data 50 associated with the matching data 50 related to the identification information 10 recognized in the previous matching process is set as the matching candidate 51, and the recognition target object 5 is used. Recognizes the identification information 10 attached to the outer surface. Therefore, in the second matching process, the matching data 50 of the second category T2 associated with the matching data 50 related to the identification information 10 recognized in the first matching process is set as the matching candidate 51, and the recognition target object 5 is used. It is a matching process for recognizing the second identification information 12 attached to the outer surface. Further, in the third matching process, the matching data 50 of the third category T3 associated with the matching data 50 related to the identification information 10 recognized in the second matching process is set as the matching candidate 51, and the recognition target object 5 is used. It is a matching process for recognizing the third identification information 13 attached to the outer surface.

In the example shown in FIG. 7, the matching data 50 (extraction matching data 52) related to the identification information 10 recognized in the first matching process is the matching data 50 associated with the information “A2”. Then, in the second matching process, the matching process uses the matching data 50 of the second category T2 (in the example shown in FIG. 7, four matching data 50) associated with the extraction matching data 52 as the matching candidate 51. Will be done. In this way, based on the recognition result of the first identification information 11 in the first matching process, the matching data 50 related to the second identification information 12 which is unlikely to be attached to the outer surface of the recognition object 5 is a matching candidate. By excluding from 51 and narrowing down the matching candidates 51, it is possible to keep the rate of erroneous recognition of the identification information belonging to the second category T2 low.

FIG. 8 is an captured image (FIG. 8) including the second identification information 12 attached to the outer surface of the recognition object 5 from among the plurality of matching data 50 included in the second category T2 by executing the second matching process. In the example shown in 6, the matching data 50 in which the degree of coincidence with the second target image G2) is equal to or greater than the threshold value is extracted. In the example shown in FIG. 8, each of the matching data 50 included in the second category T2 is the information "BN" (N = 1, 2) represented by the second identification information 12 of the generation source of each matching data 50. , ...). Then, in the example shown in FIG. 8, since the extraction matching data 52 extracted by the second matching process is associated with the information "B7", the information "B7" is recognized by the execution of the second matching process. It is recognized as the second identification information 12 attached to the outer surface of the object 5.

In the example shown in FIG. 8, the matching data 50 (extraction matching data 52) related to the identification information 10 recognized in the second matching process is the matching data 50 associated with the information “B7”. Then, in the third matching process, the matching process uses the matching data 50 of the third category T3 (three matching data 50 in the example shown in FIG. 8) associated with the extraction matching data 52 as the matching candidate 51. Will be done. In this way, based on the recognition result of the second identification information 12 in the second matching process, the matching data 50 related to the third identification information 13 which is unlikely to be attached to the outer surface of the recognition object 5 is a matching candidate. By excluding from 51 and narrowing down the matching candidates 51, it is possible to keep the rate of erroneous recognition of the identification information belonging to the third category T3 low.

FIG. 9 is an captured image (FIG. 9) including the third identification information 13 attached to the outer surface of the recognition object 5 from among the plurality of matching data 50 included in the third category T3 by executing the third matching process. In the example shown in 6, the matching data 50 in which the degree of coincidence with the third target image G3) is equal to or greater than the threshold value is extracted. In the example shown in FIG. 9, each of the matching data 50 included in the third category T3 is the information “CN” (N = 1, 2) represented by the third identification information 13 of the generation source of each matching data 50. , ...). Then, in the example shown in FIG. 9, since the extraction matching data 52 extracted by the third matching process is associated with the information "C4", the information "C4" is recognized by the execution of the third matching process. It is recognized as the third identification information 13 attached to the outer surface of the object 5.

As an example, the first category T1 is a classification of information representing the shipping source of the recognition object 5, the second category T2 is a classification of information representing the article type of the recognition object 5, and the third category T3 is. , It can be configured as a classification of information representing the property of the recognition object 5. In this case, each of the plurality of matching data 50 included in the first category T1 is associated with any of the plurality of shipping sources and stored in the storage unit 39. Further, each of the plurality of matching data 50 included in the second category T2 is associated with any of the plurality of article types, and is associated with the shipping source where the article W of the article type may be shipped. It is stored in the storage unit 39. Then, each of the plurality of matching data 50 included in the third category T3 is associated with any of the plurality of properties and is associated with an article type that may have the property, and is stored in the storage unit 39. Will be done. In this case, in the examples shown in FIGS. 7 to 9, the information "AN" (N = 1, 2, ...) Is regarded as the information representing the shipping source of the recognition object 5, and the information "BN" (N =). 1, 2, ...) Is information representing the article type of the recognition object 5, and information "CN" (N = 1, 2, ...) Is information representing the property of the recognition object 5. Will be done.

In such a configuration, the matching candidate 51 is matched with the matching data 50 associated with the item type that may be shipped from the shipping source based on the shipping source of the recognition object 5 recognized in the first matching process. Can be narrowed down to perform the second matching process. Further, based on the article type of the recognition object 5 recognized in the second matching process, the matching candidate 51 is narrowed down to the matching data 50 associated with the property that the article type may have, and the third matching process is performed. It can be carried out.

For example, the recognition object 5 can be a container (for example, a cardboard case) for accommodating agricultural products. In this case, the information indicating the shipping source is the information indicating at least one of the production area of the agricultural product contained in the recognition object 5 and the handler of the agricultural product, and the information indicating the item type is stored in the recognition object 5. The information may indicate at least one of the type of agricultural product and the product name (brand) of the agricultural product, and the information indicating the property may be information indicating at least one of the quantity and quality of the agricultural product contained in the recognition object 5. ..

The production area of agricultural products is indicated by, for example, a mark (character, graphic, photograph, picture, symbol, or a combination thereof, etc.) representing an administrative division such as a prefecture, or a mark representing an area. In addition, the handler of agricultural products is indicated by, for example, a mark representing an organization or a mark representing an organization. Specifically, the handler of agricultural products is indicated by, for example, a mark representing a cooperative (agricultural cooperative) organized by farmers. In addition, the type of agricultural product is indicated by, for example, a mark representing an item or a mark representing a variety. In addition, the quality of agricultural products is indicated by, for example, a mark indicating a grade or a mark indicating a class (size).

Here, the case where the identification information 10 is classified into the three categories T of the first category T1, the second category T2, and the third category T3 has been described as an example, but the identification information 10 is the first category T1 and Four configurations and identification information 10 classified into two categories T of the second category T2, including the first category T1, the second category T2, the third category T3, and one or more other categories T. The configuration may be classified into the above category T.

Further, one category T described here may be divided into a plurality of divided categories. For example, when the first category T1 is a classification of information representing the shipping source of the recognition target object 5, the first category T1 is a division of information representing the production area of the agricultural product contained in the recognition target object 5. It can be divided into a category and a division category which is a classification of information representing the handler of the agricultural product contained in the recognition object 5. Further, when the second category T2 is a classification of information representing the article type of the recognition target object 5, the second category T2 is a division of information representing the items of the agricultural products contained in the recognition target object 5. It can be divided into a category and a division category which is a classification of information representing the variety or product name of the agricultural product contained in the recognition object 5.

When one category T is divided into a plurality of divided categories in this way, the same processing as the above-described processing for the plurality of categories T can be performed for the plurality of divided categories in one category T. .. Specifically, the division category, which is a classification of information indicating the production area of the agricultural product contained in the recognition object 5, is set as the first division category, and the information classification indicating the handler of the agricultural product contained in the recognition object 5 is used. With a certain division category as the second division category, each of the plurality of matching data 50 included in the first division category is associated with one or more of the plurality of matching data 50 included in the second division category. It is stored in the storage unit 39, and the identification information recognition unit 32 performs a matching process for recognizing the identification information 10 belonging to the first division category attached to the outer surface of the recognition object 5, and then recognizes the identification information 10 in the matching process. The identification information 10 belonging to the second division category attached to the outer surface of the recognition object 5 with the matching data 50 of the second division category associated with the matching data 50 related to the identification information 10 as the matching candidate 51. It can be configured to perform matching processing for recognizing. In this case, each of the plurality of matching data 50 included in the first division category is associated with any of the plurality of production areas and stored in the storage unit 39. In addition, each of the plurality of matching data 50 included in the second division category is associated with any of the plurality of handlers, and is associated with the production area where the handler exists, and is stored in the storage unit 39. .. Further, the division category which is the classification of the information representing the item of the agricultural product contained in the recognition object 5 is set as the first division category, and is the classification of the information representing the variety or the product name of the agricultural product contained in the recognition object 5. When the division category is the second division category, each of the plurality of matching data 50 included in the first division category is associated with any of the plurality of items and stored in the storage unit 39. Further, each of the plurality of matching data 50 included in the second division category is associated with any of the plurality of varieties or product names, and is associated with the item in which the product type or product name exists, and is associated with the storage unit 39. Is remembered in.

[Other Embodiments]
Next, other embodiments of the article recognition system will be described.

(1) In the above embodiment, the configuration in which the article recognition system 1 (specifically, the control device 30) includes the article number recognition unit 33 has been described as an example. However, without being limited to such a configuration, if the article recognition system 1 does not recognize the number of articles W constituting the article group WG, the control device 30 does not have to include the article number recognition unit 33.

(2) In the above embodiment, the configuration in which the first imaging device 21 includes three first cameras 21a and the second imaging device 22 includes three second cameras 22a has been described as an example. However, without being limited to such a configuration, the number of the first cameras 21a included in the first imaging device 21 and the number of the second cameras 22a included in the second imaging device 22 can be appropriately changed. is there.

(3) In the above embodiment, the imaging direction changing mechanism 2 rotates the article group WG to relatively change the imaging directions of the first imaging device 21 and the second imaging device 22 with respect to the article group WG. Was described as an example. However, without being limited to such a configuration, the image pickup direction changing mechanism 2 rotates the image pickup device 20 or both the image pickup device 20 and the article group WG, so that the first item with respect to the article group WG can be rotated. The imaging direction of the imaging device 21 and the second imaging device 22 may be relatively changed. The article recognition system 1 may not be provided with the imaging direction changing mechanism 2.

(4) In the above embodiment, the first imaging device 21 and the second imaging device 22 perform the article group WG from both sides of the first horizontal direction H1, both sides of the second horizontal direction H2, and the upper side of the vertical direction V. The case where the image is configured to be imaged has been described as an example. However, the imaging direction of the article group WG by the first imaging device 21 and the second imaging device 22 can be changed as appropriate. For example, when the article number recognition unit 33 recognizes the number of articles W constituting the article group WG without being based on the image captured by the article group WG by the first imaging device 21, the article number recognition system 1 uses the article number recognition unit 33. The first image pickup apparatus 21 is configured so that the article group WG is not imaged from above the vertical direction V when the article group WG is not provided or when the articles W constituting the article group WG are arranged side by side only in the vertical direction V. May be done. Further, for example, when the identification information 10 is not attached to the upper surface S3 of the article W, or when the identification information 10 attached to the upper surface S3 of the article W is not recognized, the second imaging device 22 uses the article group WG. May be configured not to image from above the vertical direction V.

(5) In the above embodiment, in the matching process after the second matching process, the matching data 50 associated with the matching data 50 related to the identification information 10 recognized in the previous matching process is selected as the matching candidate 51. As an example, a configuration for recognizing the identification information 10 attached to the outer surface of the recognition object 5 has been described. However, the matching data 50 associated with the matching data 50 related to the identification information 10 recognized in the previous matching process in the matching process after the second matching process is not limited to such a configuration. It is also possible to perform the matching process without narrowing down the matching candidates 51. In such a configuration, unlike the above embodiment, each of the plurality of matching data 50 included in one category T is for a plurality of matchings included in a category T one lower than the category T. It may be configured to be stored in the storage unit 39 without being associated with the data 50.

(6) In the above embodiment, the configuration in which the matching data 50 of the identification information 10 belonging to each category T is stored in the storage unit 39 for each of the plurality of categories T in which the identification information 10 is classified has been described as an example. .. However, without being limited to such a configuration, the identification information 10 may not be classified into the category T, and the matching data 50 of the identification information 10 may be stored in the storage unit 39.

(7) The configurations disclosed in each of the above-described embodiments shall be applied in combination with the configurations disclosed in other embodiments as long as there is no contradiction between the embodiments described as the other embodiments. (Including combinations) is also possible. With respect to other configurations, the embodiments disclosed herein are merely exemplary in all respects. Therefore, various modifications can be made as appropriate without departing from the gist of the present disclosure.

[Outline of the above embodiment]
Hereinafter, the outline of the article recognition system described above will be described.

An article recognition system that recognizes identification information attached to the outer surface based on an image captured on the outer surface of the article, and the article group is an image group of a plurality of articles mounted on a support. Based on the images captured by the first imaging device and the second imaging device, which are two types of imaging devices, and the article group captured by the first imaging device, the captured image of the article group by the second imaging device. Is divided into a unit image which is an image of each article constituting the article group, and an identification that recognizes the identification information attached to the outer surface of the article constituting the article group based on the unit image. It is equipped with an information recognition unit.

According to this configuration, when each of the articles constituting the article group mounted on the support is to be recognized as the identification information, the identification information can be recognized based on the unit image which is an image of each article. Therefore, the identification information attached to the outer surface of each of the articles constituting the article group can be appropriately recognized.

Then, according to the above configuration, the division of the captured image of the article group into unit images by the imaging device (specifically, the second imaging device) is performed by an imaging device of a type different from that of the imaging device (specifically). Can be performed based on the captured image of the article group by the first imaging device). Unlike such a configuration, when the image captured by the image pickup device is divided into unit images based on the image captured by the image pickup device, the acquisition condition of the captured image (characteristics of the image pickup device). , Imaging conditions, etc.) must be set in consideration of both recognition of identification information and division into unit images, but acquisition conditions of captured images suitable for recognition of identification information and division into unit images It does not always match the acquisition conditions of the captured image suitable for. On the other hand, according to the above configuration, even when these two acquisition conditions are different, the acquisition condition of the image captured by the first imaging device is set as the acquisition condition suitable for division into unit images. Since the acquisition condition of the captured image by the second imaging device can be set as the acquisition condition suitable for recognizing the identification information, the identification information attached to the outer surface of each of the articles constituting the article group mounted on the support. Is easy to recognize properly.

Here, a storage unit that stores matching data related to the identification information is further provided, and the image dividing unit is used to obtain an edge image obtained by performing edge detection processing on an image captured by the first imaging device of the article group. Based on this, the image captured by the second imaging device of the article group is divided into the unit images, and the identification information recognition unit matches the unit image with the matching data stored in the storage unit. It is preferable to recognize the identification information attached to the outer surface of the articles constituting the article group by the treatment.

According to this configuration, the boundary between articles and the boundary between an article and the background can be recognized based on the edge on the edge image, and the image captured by the second imaging device of the article group can be appropriately divided into unit images. .. Then, according to the above configuration, matching data having a degree of matching with the unit image equal to or higher than the threshold value is extracted by the matching process, and the identification information attached to the outer surface of the article is recognized based on the extracted matching data. be able to.

Further, it is preferable to further include an article number recognition unit that recognizes the number of articles constituting the article group based on the captured image of the article group by the first imaging device.

According to this configuration, since the number of articles placed on the support can be recognized, it becomes easy to manage the articles on a support-by-support basis.

Further, it is preferable that the first imaging device is configured to image the article group from two or more different directions.

According to this configuration, the arrangement state of the articles constituting the article group can be recognized from two or more different viewpoints. Therefore, for example, when the directions of the articles constituting the article group are not aligned (that is, the articles of the article group). Even when the captured image contains different outer surfaces of the articles), by considering the three-dimensional shape of the articles, it becomes easy to appropriately divide the captured images of the articles into unit images. Further, when the article recognition system includes the article number recognition unit as described above, the number of articles constituting the article group is obtained by recognizing the arrangement state of the articles constituting the article group from two or more different viewpoints. Is easy to recognize with high accuracy.

Further, it is preferable to include an imaging direction changing mechanism that relatively changes the imaging direction of the first imaging device and the second imaging device with respect to the article group.

According to this configuration, while keeping the number of installations of the first imaging device and the second imaging device small, the images captured by the first imaging device and the second imaging device from two or more different viewpoints can be captured. Can be obtained. Therefore, based on the image captured by the first imaging device, the arrangement state of the articles constituting the article group can be recognized from two or more different viewpoints, and as a result, as described above, the unit of the captured image of the article group. It becomes easy to appropriately divide into images, and when the article recognition system includes an article number recognition unit, it becomes easy to recognize the number of articles constituting the article group with high accuracy. In addition, it is possible to recognize the identification information attached to the plurality of outer surfaces of the article based on the images captured by the second imaging device from two or more different viewpoints. Therefore, even if the identification information attached only to a part of the outer surface of the article exists, the identification information can be easily recognized.

The article recognition system according to the present disclosure may be capable of exerting at least one of the above-mentioned effects.

1: Article recognition system 2: Imaging direction changing mechanism 4: Support 10: Identification information 20: Imaging device 21: First imaging device 22: Second imaging device 31: Image division unit 32: Identification information recognition unit 33: Number of articles Recognition unit 39: Storage unit 50: Matching data GU: Unit image W: Article WG: Article group

Claims (5)

An article recognition system that recognizes the identification information attached to the outer surface based on the captured image of the outer surface of the article.
A first imaging device and a second imaging device, which are two types of imaging devices that image the outer surface of the article group, with the article group, which is a group of a plurality of articles mounted on the support, as an imaging target.
Based on the image of the article group captured by the first imaging device, an image dividing unit that divides the image captured by the second imaging device into a unit image that is an image of each article constituting the article group. ,
An article recognition system including an identification information recognition unit that recognizes the identification information attached to the outer surface of articles constituting the article group based on the unit image. A storage unit that stores matching data related to the identification information is further provided.
The image dividing unit converts the captured image of the article group by the second imaging device into the unit image based on the edge image obtained by performing edge detection processing on the captured image of the article group by the first imaging device. Divide and
The identification information recognition unit recognizes the identification information attached to the outer surface of the articles constituting the article group by the matching process of the unit image and the matching data stored in the storage unit. The article recognition system according to claim 1. The article recognition system according to claim 1 or 2, further comprising an article number recognition unit that recognizes the number of articles constituting the article group based on an image captured by the first image pickup device. The article recognition system according to any one of claims 1 to 3, wherein the first imaging device is configured to image the article group from two or more different directions. The article recognition system according to any one of claims 1 to 4, further comprising an imaging direction changing mechanism for relatively changing the imaging direction of the first imaging device and the second imaging device with respect to the article group.

Images