Grain inspection system

Abstract

To provide a grain inspection system capable of inspecting and confirming grains more accurately and efficiently than before.SOLUTION: A grain inspection system 100 including a measurement section 1, a server 7, and a terminal 6 connected to the server 7 via a communication line, the grain inspection system 100 including an image capturing unit 11 configured to capture an image of a grain, a feature amount calculation unit 12 configured to calculate a feature amount from the captured image of the grain captured by the image capturing unit 11, and a pseudo image generation unit 72 configured to generate a pseudo image of the grain based on the feature amount. The pseudo image generation means 72 is provided in the server 7, and the pseudo image of the grain generated by the pseudo image generation means 72 is transmitted to the terminal 6 via the communication line by the information transmission means 73 of the server 7.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a grain inspection system in which a measurement unit, a server, and a terminal are connected via a communication line. [Background technology]

[0002] As described in Patent Document 1, a system has been publicly known that is configured by connecting an imaging means for capturing images of grains, a cloud server that processes the images of the grains captured by the imaging means to determine the quality of the grains, and a terminal that displays the results of the quality determination of the grains via an internet line.

[0003] In the "grain discrimination system" described in Patent Document 1, by connecting an imaging means and a terminal prepared by a user to the cloud server via an internet line, it is possible to discriminate the quality of grains photographed by the imaging means. The cloud server is configured to store the discrimination results and photographed images of each grain, so that the user can check the discrimination results and photographed images of each grain on the terminal. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 7347629 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the conventional "grain discrimination system" described above requires that image data of grains captured by an imaging means be transmitted to a cloud server via an internet connection. Therefore, when a large number of captured images of grains with large file sizes are transmitted in real time (simultaneously), there is a risk of delays in transmission. If such a transmission delay occurs, it may become difficult to simultaneously discriminate a large number of grains, as is the case with, for example, an inline sensor (such as the "inline NIR measurement" described in Japanese Patent No. 3471009).

[0006] In view of the above problems, the present invention aims to provide a grain inspection system that can inspect and confirm grains more accurately and efficiently than conventional systems. [Means for solving the problem]

[0007] In order to solve the above problems, the present invention provides a grain inspection system comprising a measurement unit and a server, and further wherein the server and a terminal are connected to each other via a communication line, the grain inspection system comprising: an imaging means for imaging grains; a feature calculation means for calculating feature quantities from the captured image of the grains captured by the imaging means; and a pseudo image generation means for generating a pseudo image of the grains based on the feature quantities, wherein the imaging means is provided in the measurement unit, the feature calculation means is provided in the measurement unit and / or the server, the pseudo image generation means is provided in the server, and the pseudo image of the grain generated by the pseudo image generation means is transmitted to the terminal via the communication line by an information transmission means of the server.

[0008] Furthermore, in the present invention, the pseudo image generating means generates the pseudo image of the kernel by artificial intelligence using a pseudo image generation model.

[0009] Furthermore, the present invention provides a grain inspection system, wherein the feature amount is information relating to the shape and color of the grain.

[0010] Furthermore, there is provided a grain inspection system characterized by using the inline sensor of the present invention. [Effects of the Invention]

[0011] According to the present invention, since not only is large-capacity image file data transmitted to a server via a communication line as in the past, but also feature quantities calculated from a captured image can be transmitted from a measurement unit to a server according to the communication environment, it is possible to significantly reduce the transmission load and avoid delays in data transmission. In addition, since it is not necessary to store large-capacity image file data on a server, there is also the effect of reducing the load on the server for storing the image file data.

[0012] Furthermore, by providing a pseudo image generating means for generating a pseudo image of the kernel based on the above feature amount, even if the captured image of the kernel is not stored in the server, the pseudo image of the kernel captured can be sent to a terminal carried by the user, allowing the state of the kernel to be viewed and confirmed on the display device of the terminal.

[0013] Furthermore, if the pseudo image generating means generates pseudo images of kernels by artificial intelligence using a pseudo image generation model, it becomes possible to generate pseudo images of kernels with extremely high reproducibility.

[0014] Furthermore, by using an in-line sensor as the measuring unit of the present invention, in-line inspection (mass inspection) can be realized with high accuracy in various processes, not just in the receiving inspection process, in rice milling factories and the like. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a schematic configuration diagram of a grain inspection system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a schematic block diagram of a measurement unit in the present embodiment. [Figure 3] FIG. 2 is a schematic block diagram of a server according to the present embodiment. [Figure 4]FIG. 2 is a diagram illustrating an example of a configuration for calculating a feature amount according to the present embodiment. [Figure 5] 10A to 10C are diagrams illustrating a method for generating a pseudo image of a kernel in this embodiment. [Figure 6] 10 is a table showing an example of calculation of feature amounts in the present embodiment. [Figure 7] FIG. 2 is a diagram illustrating an example of a pseudo image generation model in the present embodiment. [Figure 8] FIG. 10 is a diagram showing an example of a pseudo image generated in this embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, an embodiment of the grain inspection system of the present invention will be described with reference to the drawings. However, the present invention is not limited to the embodiment shown below.

[0017] 1 shows a schematic configuration diagram of a grain inspection system 100 according to this embodiment. As shown in the figure, the grain inspection system 100 includes at least a measurement unit 1, a server 7, and a terminal 6.

[0018] The measurement unit 1 of this embodiment is an inline sensor for rice milling facilities or grain processing facilities, and is applied to devices that can continuously capture images of grains transported by conveying means such as chutes or belts, such as known grain discrimination devices and color sorters equipped with imaging equipment such as optical cameras.

[0019] Of course, the measurement unit 1 of the present invention is not limited to the above-mentioned inline sensor, but can also be applied as an online sensor to a grain inspection device for online analysis of grains collected via a sampling device.

[0020] Furthermore, the measurement unit 1 of the present invention can also be applied as an offline sensor to a tabletop inspection device or the like for offline analysis of collected grains in a laboratory.

[0021] The measurement unit 1 is connected to the server 7 via a communication line, and is capable of transmitting grain feature quantities, quality determination results, and the like, which will be described later, to the server 7 in response to various commands from the server 7. The configuration of the communication line and the server 7 is not particularly limited, but an internet line can be suitably used as the communication line. Furthermore, as the server 7, it is possible to use, for example, a regular server installed in the business premises where the terminal 6 is used or in a related company, but a cloud server can also be suitably used.

[0022] The server 7 and the terminal 6 are connected by a communication line in the same manner as above, and in response to various commands from the terminal 6, it is possible to send pseudo images of grains and quality determination results, which will be described later, from the server 7 to the terminal 6. The communication line is not particularly limited, but an internet line can be suitably used as the communication line.

[0023] The terminal 6 may be a smartphone owned by the user, a PDA, a tablet terminal, a notebook PC, a desktop PC, etc. In particular, a device with excellent portability is preferably used.

[0024] Furthermore, as an additional function, the measurement unit 1 may be configured to be linked to the terminal 6. That is, the measurement unit 1 has unique identification information, and by inputting the unique identification information of the measurement unit 1 into the terminal 6, the measurement unit 1 can be linked to the terminal 6. This allows information such as the discrimination results and pseudo images of the kernels corresponding to the measurement unit 1 to be efficiently viewed and confirmed by the terminal 6.

[0025] The unique identification information may be input to the terminal 6 by reading a two-dimensional code displayed on the measurement unit 1 using a camera mounted on the terminal 6, or by keying in an ID individually assigned to the measurement unit 1 into the terminal 6. In addition, a selection may be made from a plurality of pre-registered measurement units 1 displayed on the terminal 6.

[0026] Next, a schematic block diagram of the measurement unit 1 in this embodiment is shown in Figure 2. As shown in the figure, the measurement unit 1 includes at least an imaging means 11 that captures an image of a grain to be inspected, a feature amount calculation means 12 that calculates a feature amount from the image of the grain captured by the imaging means 11, and a transmission means 14 that transmits the feature amount calculated by the feature amount calculation means 12 to the server 7 via a communication line.

[0027] In this embodiment, a quality discrimination means 13 is further provided that can discriminate the quality of the grain, and it is possible to discriminate the quality of the grain based on the image of the grain captured by the imaging means 11 and / or the feature calculated by the feature calculation means 12.

[0028] In this embodiment, the imaging means 11 is equipped with an optical camera (not shown) that captures images of the multiple grains flowing down. In addition, multiple light sources that emit light toward the grains are provided, and white LEDs are provided. Note that the light sources are not limited to white LEDs, and it is also possible to use single-color LEDs of R (red), G (green), and B (blue).

[0029] In addition, the feature amounts in this embodiment are information related to the shape and color of the grain to be inspected. Explaining in more detail, as shown in Fig. 4, feature amount calculation means 12 calculates the feature amounts of the grain from an image of a grain 2 captured by imaging means 11, such as grains 1 to 3. For example, as shown in the table in Fig. 4, it is possible to calculate the "kernel length," "kernel width," and "area" of each grain (whole grain, broken grain, partially colored grain) as feature amounts.

[0030] Note that the feature quantities are not limited to the above-mentioned "grain length," "grain width," and "area." For example, as shown in the table in FIG. 6, in addition to the "length," "width," and "area" of the grain, "perimeter" and "center of gravity" can also be calculated as feature quantities for the shape characteristics of the entire grain. Furthermore, it is possible to calculate the color characteristics of the entire grain, the shape characteristics of the colored region, and the color characteristics of the colored region as feature quantities. These feature quantities are then transmitted to the server 7 via a communication line. This configuration eliminates the need to transmit large amounts of image file data to the server 7 as in the past, significantly reducing the volume of transmitted data and significantly reducing the transmission load compared to the past.

[0031] As shown in Figure 3, the server 7 of this embodiment is equipped with at least a pseudo image generation means 72 that generates a pseudo image of the kernel based on the above-mentioned feature values ​​transmitted by the transmission means 14 of the measurement unit 1, and an information transmission means 73 that transmits the pseudo image of the kernel generated by the pseudo image generation means 72 to the terminal 6 via a communication line.

[0032] The server 7 of this embodiment further includes a storage means 71, which is capable of storing the above-mentioned feature amounts transmitted by the transmission means 14 of the measurement unit 1, the results of kernel quality determination by the quality determination means 13, etc. It is also possible to configure the storage means 71 to store the pseudo image of the kernel generated by the pseudo image generation means 72.

[0033] 5, when the user operates the terminal 6 to request the server 7 to display an image of the grain to be inspected, the server 7 reads out the feature quantities of the grain corresponding to a predetermined inspection date and lot number from the storage means 71. Then, based on the read-out feature quantities, a pseudo image of the corresponding grain is generated by the pseudo image generation model of the pseudo image generation means 72. The generated pseudo image of the grain is transmitted from the information transmission means 73 to the terminal 6 via a communication line, and the user can view and confirm the pseudo image of the grain via the display device of the terminal 6.

[0034] In this embodiment, the pseudo-image generating means 72 generates pseudo-images of grains using artificial intelligence (AI) with a pseudo-image generation model. For example, as shown in FIG. 7, a large number of sets of images of whole grains and their feature values, images of broken grains and their feature values, and images of partially colored grains and their feature values ​​can be prepared as training data to construct a pseudo-image generation model. It is possible to generate highly accurate pseudo-images of grains other than whole grains, broken grains, and partially colored grains by increasing the feature values ​​and grain types shown in the table of FIG. 6 and training the model. For example, for brown rice, pseudo-images of milky white grains, immature grains at the base, white grains at the back and ventral sides, immature green grains, dead green grains, dead white grains, fully colored grains, partially colored grains, and cracked grains can be generated. For polished rice (white rice), pseudo-images of cracked grains, fully colored grains, partially colored grains, damaged grains, fully chalky grains, semi-chalky grains, and white grains at the back and ventral sides can be generated.

[0035] Furthermore, as shown in FIG. 8, it is possible to generate pseudo images with high reproducibility of various grains such as "short grain," "medium grain," and "long grain."

[0036] (Other variations) Although one embodiment of the grain inspection system of the present invention has been described above, the present invention is not limited to the above embodiment and also includes the following.

[0037] The grains that can be inspected by the grain inspection system include granular materials such as rice, wheat, beans, and corn.

[0038] Furthermore, it is possible to use a generative AI consisting of an autoencoder as a pseudo-image generation model, and it is also possible to add a configuration that allows the system to find features by itself through learning. This makes it possible to generate pseudo-images of grains with even higher accuracy.

[0039] In the above-described embodiment, the case where the feature calculation means 12 and the quality discrimination means 13 are provided on the measurement unit 1 side under conditions where communication delays are not acceptable has been described. However, for example, in cases where there is sufficient capacity for data transmission via a communication line, it is also possible to provide the feature calculation means 12 and the quality discrimination means 13 on the server 7, and to have the server 7 calculate the feature values ​​of the kernels.

[0040] Specifically, after the image of the kernel photographed by the measurement unit 1 is received by the server 7, feature amounts are calculated from the photographed image and the feature amounts are stored in the server 7. Therefore, once the feature amounts have been stored in the server 7, there is no need to store the photographed image in the server 7. Therefore, there is no need to provide the server 7 with a large capacity memory. Furthermore, there is no need to provide the measurement unit 1 with advanced and expensive computing capabilities, which is expected to result in an effect of allowing the measurement unit 1 to be configured inexpensively.

[0041] In yet another embodiment, the feature calculation means 12 and / or the quality discrimination means 13 can be provided on both the measurement unit 1 side and the server 7 side. With this configuration, it becomes possible to select whether to send the image of the kernel photographed by the measurement unit 1 to the server 7 and calculate the feature on the server 7 side, or to calculate the feature on the measurement unit 1 side without sending the image of the kernel photographed by the measurement unit 1 to the server 7, depending on the communication environment (presence or absence of communication delay and its degree).

[0042] In addition, in the above-described embodiment, the measurement unit 1 is configured to be equipped with a quality discrimination means 13 capable of discriminating the quality of grains, but this is not necessarily limited to this, and it is also possible to provide the quality discrimination means 13 in the server 7.

[0043] Although several embodiments of the present invention have been described above, the above-described embodiments of the invention are intended to facilitate understanding of the present invention and are not intended to limit the present invention. The present invention may be modified or improved without departing from its spirit, and the present invention includes equivalents thereof. Furthermore, the components described in the claims and specification may be combined or omitted to the extent that at least some of the above-described problems can be solved or at least some of the effects can be achieved. [Explanation of symbols]

[0044] 1 Measuring part 11 Imaging means 12 Feature calculation method 13 Quality determination means 14 Transmission Method 2 grains 6 Terminals 7 Server 71 Memory means 72 Pseudo image generation means 73 Means of information transmission 100 Grain Inspection System

Claims

1. In a grain inspection system comprising a measurement unit and a server, and further comprising the server and a terminal connected to each other via a communication line, imaging means for imaging the grain; a feature amount calculation means for calculating feature amounts from the captured image of the kernels captured by the imaging means; a pseudo image generating means for generating a pseudo image of the kernel based on the feature amount, the imaging means is provided in the measurement unit, the feature calculation means is provided in the measurement unit and / or the server, The pseudo image generating means is provided in the server, The pseudo image of the kernel generated by the pseudo image generating means is transmitted to the terminal via the communication line by the information transmitting means of the server. A grain inspection system characterized by:

2. The pseudo-image generating means generates the pseudo-image of the kernel by artificial intelligence using a pseudo-image generation model. The grain inspection system according to claim 1 .

3. The feature amount is information relating to the shape and color of the kernel.

3. The grain inspection system according to claim 1 or 2.

4. The measuring unit is an in-line sensor 3. The grain inspection system according to claim 1 or 2.

5. The measuring unit is an in-line sensor 4. The grain inspection system according to claim 3.

Images