JP7085952B2 - Change rate estimation device, change rate estimation method, state estimation device and state estimation method - Google Patents

Description

The present invention relates to a rate of change estimation device and a method of estimating the rate of change for estimating information about the target rice, and also calculates the specific gravity, density or water absorption rate of the target rice based on the estimated information to estimate the state of the target rice. The present invention relates to a state estimation device and a state estimation method.

For example, in sake brewing, the processing process for raw rice such as the rice milling process, the water absorption process, and the steaming process is an important process that influences the subsequent brewing process. In the water absorption process when manufacturing Daiginjo Sake, etc., there is a case where the so-called limited water absorption is performed by shortening the soaking time so that the raw rice does not absorb too much water and limiting the water absorption of the raw rice. Since it is required to adjust the water absorption rate in units of 1%, it is extremely important to grasp the water absorption state of the raw rice in real time in order to improve the quality of the liquor produced.

It is known that the change in water absorption of raw rice varies greatly depending on the condition of the raw rice (variety, rice polishing rate, water content, etc.) and immersion conditions (time, temperature, concentration, etc.). As a method for evaluating the change in water absorption of rice, a method of measuring the weight of raw rice before and after immersion with a weighing machine and calculating the amount of water absorption from the difference in weight before and after immersion was common. However, with such a method, it is not possible to accurately grasp the change over time in the amount of water absorption and the distribution of water absorption. Therefore, at the actual brewing site, the water absorption distribution is estimated visually from the change in the color and shape of the raw rice, and the water absorption change is estimated together with the above-mentioned calculation result of the water absorption amount, and the water absorption state of the raw rice is objectively viewed. There was a need to develop a method that could be evaluated in real time.

By the way, the applicant of the present application has proposed the method for determining the water content state described in Patent Document 1 as a method for determining the water absorption amount and the water absorption distribution of the rice to be cooked by observing an image. According to this water content determination method, the rice to be cooked placed in the state of being immersed in water can be optically monitored, and the water content-related information regarding the water content of the rice to be cooked can be measured over time. , It is possible to evaluate the water absorption amount and water absorption distribution of rice to be cooked based on the water content related information. Specifically, the change in the size of the rice to be cooked due to water absorption of the rice to be cooked is measured over time, and the size of the rice to be cooked after water absorption is compared with the size of the rice to be cooked before water absorption. Since the rate of change (expansion) can be used as the expansion rate, the water absorption amount and water absorption distribution of the rice to be cooked can be evaluated based on this expansion rate.

Japanese Unexamined Patent Publication No. 2016-105074

According to the water content determination method described in Patent Document 1, it is possible to evaluate the water absorption amount and the water absorption distribution (water absorption state) of the rice to be cooked. However, the water absorption state of rice varies greatly depending on the type of rice, and the rice to be cooked expands mainly in the direction perpendicular to the vertical direction due to water absorption, whereas the rice to be brewed is orthogonal to the vertical direction due to water absorption. It expands not only in the direction of rice, but also in the vertical direction. Therefore, when the water-containing state determination method described in Patent Document 1 is used, for example, simply by optically monitoring the rice to be cooked placed in a state of being immersed in water from above in the vertical direction, the direction orthogonal to the vertical direction is used. Although it is possible to know the degree of expansion to, it is not possible to know the degree of expansion in the vertical direction. Therefore, when the water content of rice that expands in the vertical direction (for example, rice to be brewed) is evaluated using the water content determination method described in Patent Document 1, the water content-related information reflects the expansion in the vertical direction. Therefore, there is a discrepancy between the measured water content-related information and the actual water content of the sake brewing target rice, and as a result, the water absorption state of the sake brewing target rice cannot be accurately evaluated.

In other words, in order to accurately evaluate the water absorption state of rice regardless of the type of rice, we considered only the rate of change in shape of the rice to be evaluated in the two-dimensional direction (rate of change in two-dimensional shape). Is insufficient, and it is important to consider the rate of change in shape in the direction orthogonal to the two-dimensional direction and the rate of change in shape in the three-dimensional direction (rate of change in three-dimensional shape).

If the rice is optically monitored from the direction orthogonal to the vertical direction by using the water-containing state determination method described in Patent Document 1, the rate of change of the shape in the vertical direction and the rate of change of the three-dimensional shape can be obtained. In this case, a device for optically monitoring the rice (a lighting device that irradiates the target rice, a camera device that captures the transmitted light from the target rice, etc.) is provided for each direction. Since it is necessary to provide the equipment in the above, there is a concern that the equipment configuration becomes complicated and the equipment cost increases.

Further, the rate of change in shape and the rate of change in three-dimensional shape in a direction orthogonal to the two-dimensional direction can be used not only for evaluation of the water absorption state of rice, but also in a direction orthogonal to the two-dimensional direction (thickness of rice), for example. If the specific gravity of rice is calculated based on the rate of change of the shape in the direction) and the rate of change of the three-dimensional shape, it is possible to evaluate the state of the white heart, and the shape in the direction orthogonal to the two-dimensional direction of the rice. There are many advantages to finding the rate of change in rice and the rate of change in the three-dimensional shape of rice, and there is also a need to develop a method that can easily find these.

The present invention has been made in view of the above circumstances, and it is possible to easily obtain the rate of change in shape in the thickness direction of rice without requiring a device having a complicated configuration, or to correct the rate of change in the two-dimensional shape of rice. By providing a change rate estimation device and a change rate estimation method that can easily obtain the change rate of the three-dimensional shape, the change rate of the shape and the change rate of the three-dimensional shape in the direction orthogonal to the two-dimensional direction are provided. It is an object of the present invention to provide a state estimation device and a state estimation method capable of easily estimating the specific gravity, density or water absorption rate of rice based on the above.

The characteristic configuration of the rate of change estimation device according to the present invention for achieving the above object is
Photographing means for photographing the target rice in a state of being immersed in water,
An image data acquisition means for acquiring two-dimensional image data of the target rice photographed by the photographing means and viewing the target rice from the thickness direction.
With reference to the two-dimensional image data acquired by the image data acquisition means, the two-dimensional image data is analyzed, and among a plurality of target rice, a vertical crack in which a vertical cross section along the long axis direction of the target rice is opened. A crack rate calculation means for calculating the vertical crack rate, which is the ratio of the target rice in which
The vertical cracking rate is obtained from the cracking rate calculating means, and the target is based on the vertical cracking rate and the predetermined correlation between the vertical cracking rate of the target rice and the rate of change in shape in the thickness direction. The point is that it is provided with an estimation means for estimating the rate of change in the shape of rice in the thickness direction.
In addition, the characteristic configuration of the rate of change estimation method according to the present invention for achieving the above object is
The shooting process of shooting the target rice soaked in water,
Based on the two-dimensional image of the target rice taken and viewed from the thickness direction, vertical cracks with an open vertical cross section along the long axis direction of the target rice occur among a plurality of target rice. The crack rate calculation process that calculates the vertical crack rate, which is the ratio of the target rice,
Based on the calculated vertical cracking rate and the predetermined correlation between the vertical cracking rate of the target rice and the shape change rate in the thickness direction, the change rate of the shape of the target rice in the thickness direction is determined. The point is to execute the estimation process to be estimated.

Here, in the process of studying a method for determining the water absorption rate and water absorption distribution of rice by observing an image, the inventor of the present application differs in the cracking method of rice in a state of being immersed in water depending on the rice, and the water absorption rate depends on the cracking method of rice. It was discovered that there is a difference in the state of water absorption, in other words, the rate of change in the three-dimensional shape of rice. As a method of determining the cracking method, there is a method of visually determining the cracking method, but it is difficult to objectively record the cracking method of rice, which changes from moment to moment, as a numerical value. Therefore, the inventors of the present application repeated diligent research, and as a result, succeeded in quantifying how the rice cracked by performing automatic analysis using artificial intelligence on the two-dimensional image data, and further, vertical among the target rice. We obtained new findings that the proportion of target rice with cracks correlates with the rate of change in shape of the target rice in the thickness direction.

The rate of change estimation device and the method for estimating the rate of change according to the present invention were made based on such findings. In the above characteristic configuration, first, the target rice in a state of being immersed in water was viewed from the thickness direction 2 The dimensional image is photographed by the photographing means, and then the two-dimensional image data photographed by the photographing means is acquired by the image data acquisition means. Next, the crack rate calculating means analyzes the two-dimensional image data with reference to the two-dimensional image data, and determines the vertical crack rate, which is the ratio of the target rice in which vertical cracks occur among the plurality of target rice. calculate.

After that, the estimation means obtains the vertical cracking rate, and based on the vertical cracking rate and the correlation between the predetermined vertical cracking rate of the target rice and the rate of change in shape in the thickness direction, the target rice is used. Estimate the rate of change in shape in the thickness direction.

As described above, according to the above characteristic configuration, the vertical cracking rate having a correlation with the rate of change in shape in the thickness direction of the target rice is calculated, and the vertical cracking rate and the vertical cracking rate in the thickness direction of the target rice are based on the above correlation. Since the rate of change in shape in the rice can be estimated, the rate of change in shape of the target rice in the thickness direction can be easily obtained without optically monitoring the target rice from the direction orthogonal to the thickness direction as in the conventional case. be able to. Therefore, it becomes possible to evaluate, for example, the water absorption state of the target rice and the state of whiteness of the target rice based on the obtained rate of change in shape in the thickness direction.

The characteristic configuration of the state estimation device according to the present invention for achieving the above object is
With the above rate of change estimation device,
A point including a state calculation means for acquiring the rate of change in shape in the thickness direction from the estimation means and calculating the specific gravity, density or water absorption rate of the target rice based on the acquired rate of change in shape in the thickness direction. It is in.
Further, the characteristic configuration of the state estimation method according to the present invention for achieving the above object is
After performing the change rate estimation method, a state calculation step of calculating the specific gravity, density or water absorption rate of the target rice based on the change rate of the shape in the thickness direction estimated by the change rate estimation method is executed. It is in.

According to the above-mentioned feature configuration, it is possible to calculate the specific gravity, density or water absorption rate of the target rice based on the rate of change in shape in the direction orthogonal to the thickness direction, which is easily obtained without the need for an optical monitoring device. Therefore, the state of the target rice such as specific gravity, density, and water absorption rate can be estimated more easily than before.

Further, a further characteristic configuration of the state estimation device according to the present invention is
Area information acquisition that analyzes the two-dimensional image data with reference to the two-dimensional image data acquired by the image data acquisition means and acquires the area of the target rice in a direction along a plane orthogonal to the thickness direction. Equipped with means,
In the state calculation means, the area is acquired from the area information acquisition means, the rate of change in shape in the acquired thickness direction, the acquired area, and the thickness of the target rice before being immersed in pre-stored water. The point is to calculate the specific gravity or density of the target rice based on the volume of the target rice calculated based on the above.
Further, the further characteristic configuration of the state estimation method according to the present invention is
The area information acquisition step for acquiring the area of the target rice in the direction along the plane orthogonal to the thickness direction is provided based on the two-dimensional image of the target rice taken and viewed from the thickness direction. ,
In the state calculation step, the target rice is calculated based on the rate of change in shape in the thickness direction, the acquired area, and the volume of the target rice before being immersed in water. The point is to calculate the specific gravity or density of.

According to the above feature configuration, the area of the target rice in the direction along the plane orthogonal to the thickness direction is acquired based on the two-dimensional image data, and the acquired area and the shape in the thickness direction more easily obtained than before are obtained. In addition to the rate of change of, the thickness of the target rice before soaking in water is obtained in advance, the volume is calculated based on these, and the specific gravity or density of the target rice is calculated based on this volume. .. Therefore, the state of the specific weight or density of the target rice can be estimated more easily than before.

The characteristic configuration of the rate of change estimation device according to the present invention for achieving the above object is
Photographing means for photographing the target rice in a state of being immersed in water,
An image data acquisition means for acquiring two-dimensional image data of the target rice photographed by the photographing means and viewing the target rice from the thickness direction.
With reference to the two-dimensional image data acquired by the image data acquisition means, the two-dimensional image data is analyzed, and the rate of change of the two-dimensional shape of the target rice in the direction along the plane orthogonal to the thickness direction is determined. Two-dimensional shape change rate calculation means to calculate,
The change rate of the two-dimensional shape is acquired from the two-dimensional shape change rate calculation means, and among a plurality of target rice, the target rice in which a vertical crack having a vertical cross section open along the long axis direction of the target rice is generated. Based on the correlation between the vertical cracking rate, which is the ratio of The point is that it is provided with an estimation means for estimating the rate of change.
Further, the characteristic configuration of the rate of change estimation method in the present invention for achieving the above object is
The shooting process of shooting the target rice soaked in water,
Based on the two-dimensional image of the target rice taken and viewed from the thickness direction, the rate of change in the two-dimensional shape of the target rice in the direction along the plane orthogonal to the thickness direction is calculated. 2 Dimensional shape change rate calculation process and
Correlation between the vertical cracking rate, which is the ratio of the target rice having vertical cracks with an open vertical cross section along the long axis direction of the target rice, and the rate of change in shape in the thickness direction. , And an estimation step of correcting the change rate of the two-dimensional shape and estimating it as the change rate of the three-dimensional shape based on the calculated change rate of the two-dimensional shape.

As described above, the inventors of the present application have different rates of change in the three-dimensional shape of rice depending on how the rice is cracked, and the proportion of rice in which vertical cracks occur is in the thickness direction of the rice. We obtained a new finding that there is a correlation with the rate of change in shape in.

The change rate estimation device and the change rate estimation method according to the present invention were made based on such findings. In the above configuration, first, a two-dimensional image of the target rice immersed in water seen from the thickness direction. Is photographed by the photographing means, and then the image data acquisition means acquires the two-dimensional image data photographed by the photographing means. Next, the two-dimensional shape change rate calculation means analyzes the two-dimensional image data with reference to the two-dimensional image data, and determines the change rate of the two-dimensional shape of the target rice in the direction along the plane orthogonal to the thickness direction. calculate.

After that, the estimation means acquires the rate of change of the two-dimensional shape, and based on the correlation between the rate of vertical cracking and the rate of change of the shape in the thickness direction, and the rate of change of the acquired two-dimensional shape, the two-dimensional shape. The rate of change of the two-dimensional shape calculated by the rate of change calculation means is estimated as the rate of change of the three-dimensional shape.

As described above, according to the above-mentioned feature configuration, the two-dimensional shape change rate can be corrected to obtain the three-dimensional shape change rate based on the above-mentioned correlation and the acquired two-dimensional shape change rate. Regardless of the type of the target rice, the rate of change in the three-dimensional shape of the target rice can be easily estimated. Therefore, based on the estimated rate of change of the three-dimensional shape, for example, the water absorption state of the target rice and the state of the white heart of the target rice can be evaluated.

A further characteristic configuration of the rate of change estimation device according to the present invention is that in the estimation means, the degree of influence of vertical cracks determined based on the correlation is multiplied by the rate of change of the acquired two-dimensional shape to obtain the two-dimensional shape. The point is to correct the rate of change to the rate of change of the three-dimensional shape.
Further, in the further characteristic configuration of the change rate estimation method according to the present invention, the degree of influence of vertical cracks determined based on the correlation in the estimation step is multiplied by the change rate of the calculated two-dimensional shape to obtain the above 2 The point is to correct the rate of change of the three-dimensional shape to the rate of change of the three-dimensional shape.

Here, the "degree of influence of vertical cracks" is an index showing the influence of the presence of vertical cracks on the rate of change of the three-dimensional shape, and specifically, a database obtained from measurement results of a plurality of varieties in advance. A graph showing the correlation between the vertical crack rate and the rate of change in the thickness direction is created based on the above, and the slope of this graph is determined in advance as the vertical crack influence degree.

According to the above characteristic configuration, by multiplying the rate of change of the two-dimensional shape of the target rice by the degree of influence of vertical cracking, the rate of change of the two-dimensional shape can be corrected to obtain the rate of change of the three-dimensional shape of the target rice. can. Therefore, similarly to the above, the rate of change in the three-dimensional shape of the target rice can be easily obtained, and the water absorption state of the rice and the state of the white heart of the rice can be evaluated.

Further, the characteristic configuration of the state estimation device according to the present invention for achieving the above object is the change rate estimation device according to any one of the above.
The point is to provide a state calculation means for acquiring the rate of change of the three-dimensional shape from the estimation means and calculating the specific gravity, density or water absorption rate of the target rice based on the acquired rate of change of the three-dimensional shape. ..
Further, the characteristic configuration of the state estimation method according to the present invention for achieving the above object is to perform the change rate of the three-dimensional shape estimated by the change rate estimation method after performing any of the change rate estimation methods. Based on this, the point is to execute the state calculation step of calculating the specific gravity, density or water absorption rate of the target rice.

According to the above feature configuration, the specific gravity of the target rice is based on the change rate of the three-dimensional shape obtained by correcting the change rate of the two-dimensional shape, that is, the change rate of the three-dimensional shape obtained more easily than before. Since the density or the water absorption rate can be calculated, it is possible to estimate the state of the target rice such as the specific gravity, the density, and the water absorption rate more easily than before.

Further characteristic configurations of the state estimation device according to the present invention include the rate of change of the acquired three-dimensional shape in the state calculation means, and the thickness and thickness direction of the target rice before being immersed in water stored in advance. The point is to calculate the specific gravity or density of the target rice based on the volume of the target rice calculated based on the area in the direction along the orthogonal plane.
Further, a further characteristic configuration of the state estimation method according to the present invention is the rate of change of the three-dimensional shape in the state calculation step, and the thickness and thickness direction of the target rice before being immersed in water stored in advance. The point is to calculate the specific gravity or density of the target rice based on the volume of the target rice calculated based on the area in the direction along the orthogonal plane.

According to the above feature configuration, in addition to the rate of change of the three-dimensional shape that can be easily obtained than before, the thickness of the target rice before being immersed in water and the area in the direction along the plane orthogonal to the thickness direction are obtained in advance. Then, the volume is calculated based on these, and the specific gravity or density of the target rice is calculated based on this volume. Therefore, the state of the specific weight and density of the target rice can be estimated more easily than before.

In the present application, the "specific gravity" means the specific gravity (cm ³ / g) as the reciprocal of the density (g / cm ³ ).

It is a figure which shows the structure of the state estimation apparatus and change rate estimation apparatus which concerns on 1st Embodiment. It is a flowchart explaining the process of the state estimation method and the change rate estimation method. It is a figure explaining the example of image processing. It is a figure which shows the example of the form of sake rice. It is a graph which shows the relationship between the change rate of the two-dimensional shape and the change rate of the three-dimensional shape of sake rice, and the water absorption rate. It is a graph which shows the relationship between the vertical crack rate and the thickness change rate. It is a flowchart explaining a machine learning process. It is a graph which shows the relationship between the number of learnings and the correct answer rate. It is a graph which shows the relationship between the number of learnings and the correct answer rate. It is a graph which shows the relationship between the number of learnings and the correct answer rate. It is a figure which shows the structure of the state estimation apparatus and change rate estimation apparatus which concerns on 2nd Embodiment. It is a flowchart explaining a machine learning process. It is a flowchart explaining the validity determination process. It is a figure which shows the structure of the state estimation apparatus and change rate estimation apparatus which concerns on 3rd Embodiment. It is a flowchart explaining the process of the state estimation method and the change rate estimation method.

[First Embodiment]
Hereinafter, the state estimation device 1 and the rate of change estimation device 20A according to the embodiment of the present invention will be described with reference to the drawings. In this embodiment, a case of estimating the rate of change in shape of sake rice in the thickness direction will be described as an example.

As shown in FIG. 1, the state estimation device 1A (1) according to the present embodiment includes a photographing device (shooting means) 2, an image data acquisition unit (image data acquisition means) 11, an image processing unit 12, and an area. Information to be handled, including an information acquisition unit (area information acquisition means) 13, a crack rate calculation unit (crack rate calculation means) 16, an estimation unit (estimation means) 19, and a state calculation unit (state calculation means) 30. A storage device 35 for storing the data is provided. In FIG. 1, the state estimation device 1 includes a change rate estimation device 20A (20) including a photographing device 2, an image data acquisition unit 11, an image processing unit 12, a crack rate calculation unit 16, an estimation unit 19, and a storage device 35. It is configured to include an area information acquisition unit 13 and a state calculation unit 30, and also includes an image data acquisition unit 11, an image processing unit 12, an area information acquisition unit 13, a crack rate calculation unit 16, and an estimation unit 19. The state calculation unit 30 and the storage device 35 are illustrated in a form constituting the rice analyzer 10A (10).

The US analyzer 10A, which constitutes a part of the state estimation device 1, is realized by using one or a plurality of computer devices having an information calculation processing function, an information input / output function, an information storage function, and the like. .. In that case, the function of the image data acquisition unit 11, the function of the image processing unit 12, the function of the area information acquisition unit 13, and the crack rate calculation unit 16 (the function of the morphology determination unit 17, the function of the morphology data analysis unit 18). The function of the above, the function of the estimation unit 19, and the function of the state calculation unit 30 may be installed in the computer device.

The storage device 35 of the present embodiment includes an image data storage unit 35a for storing image data captured by the imaging device 2, an area storage unit 35b for storing area information acquired by the area information acquisition unit 13, and machine learning. It includes a training data storage unit 35c for storing training data to be used, and a morphological storage unit 35d for storing the determination result of the morphological determination unit 17. Here, the area storage unit 35b and the morphology storage unit 35d each use the area information acquired by the area information acquisition unit 13 and the determination result of the morphology determination unit 17 as one target rice after starting immersion in water. It may be stored in a state where the change history of the area and morphology of the rice can be known. Further, the storage device 35 of the present embodiment stores in advance the thickness of the sake rice before it is immersed in water.

The photographing device 2 is arranged vertically above the rice holding portion in which the sake rice is immersed in water, and captures the transmitted light from the sake rice to capture a two-dimensional image of the sake rice. It is a device that can be used.

The output device 3 is a display device capable of displaying image information, character information, and the like, and a printing device capable of printing the image information, character information, and the like on paper or the like.

FIG. 2 shows a step of acquiring the area information of sake rice (area information acquisition step) and a step of calculating the cracking rate of sake rice (breaking rate calculation) included in the change rate estimation method and the state estimation method according to the present embodiment. Step), a step of estimating the change in shape of sake rice in the thickness direction (estimation step), and a step of calculating the specific weight of sake rice based on the estimated rate of change in shape in the thickness direction (state calculation step) are explained. It is a flowchart to be done. After the step # 10 of FIG. 2 is performed, that is, after the shooting step of shooting the sake rice placed in a state of being immersed in water by the shooting device 2 is executed, in the step # 11, the image data acquisition unit 11 Acquires two-dimensional image data of sake rice placed in a state of being immersed in water taken by the photographing apparatus 2 from the thickness direction of the sake rice. That is, an image data acquisition step of acquiring two-dimensional image data of sake rice placed in a state of being immersed in water as viewed from the thickness direction of the rice is executed. For example, the photographing device 2 photographs a plurality of sake rice arranged in a state of being immersed in water to obtain one two-dimensional image data. Then, the two-dimensional image data is transmitted to the rice analyzer 10A and acquired by the image data acquisition unit 11. The two-dimensional image data acquired by the image data acquisition unit 11 can also be stored in the image data storage unit 35a of the storage device 35. For example, the image data A shown in FIG. 3 is two-dimensional image data acquired by the image data acquisition unit 11.

Next, in step # 12 of FIG. 2, the image processing unit 12 acquires image data prior to the process of acquiring area information by the area information acquisition unit 13 and the process of calculating the vertical crack rate by the crack rate calculation unit 16. In the two-dimensional image data acquired by the unit 11, image processing including a process of extracting the two-dimensional liquor rice image data of a region containing only one liquor rice is performed, and the processed two-dimensional image of the one liquor rice is performed. Generate data. That is, in the two-dimensional image data acquired in the image data acquisition step, image processing including a process of extracting the two-dimensional liquor rice image data of the region containing only one liquor rice is performed, and the one liquor rice has been processed. The image processing step of generating the two-dimensional image data of the above is executed. FIG. 3 is a diagram illustrating an example of image processing. In the present embodiment, as the image processing performed by the image processing unit 12, one of the processing of extracting the two-dimensional liquor rice image data of the region containing only one liquor rice from the image data acquired by the image data acquisition unit 11. It includes a process of rotating a two-dimensional image of sake rice so that the sake rice is in a predetermined posture, and a process of adjusting the contrast of the image.

For example, the two-dimensional image data B shown in FIG. 3 is an image obtained by extracting data including a two-dimensional whole image of one sake rice from the image data A acquired by the image data acquisition unit 11. The two-dimensional image data C shown in FIG. 3 is a two-dimensional liquor in a region containing only one liquor rice by removing a two-dimensional image of other liquor rice other than one liquor rice in the two-dimensional image data B. This is an image after processing for extracting rice image data. The two-dimensional image data D shown in FIG. 3 is an image after processing for rotating a two-dimensional image of sake rice so that one sake rice has a predetermined posture with respect to the two-dimensional image data C. .. In this case, the sake rice image is rotated so that the long axis of the sake rice is along the vertical direction. The two-dimensional image data E shown in FIG. 3 is an image after performing a process of adjusting the orientation of sake rice (for example, a vertical inversion process, a left-right inversion process, etc.) on the two-dimensional image data. The two-dimensional image data F shown in FIG. 3 is an image after processing for adjusting the contrast of the image with respect to the two-dimensional image data E.

Next, the area information acquisition unit 13 analyzes the two-dimensional image data with reference to the two-dimensional image data acquired by the image data acquisition unit 11, and the liquor in the direction along the plane orthogonal to the thickness direction of the liquor rice. Calculate the area of rice.

In the present embodiment, the area information acquisition unit 13 refers to the two-dimensional image data acquired by the image data acquisition unit 11 and processed by the image processing unit 12, and the direction along the plane orthogonal to the thickness direction of the sake rice. The area of sake rice in the above can be automatically determined using various image recognition techniques.

As described above, the two-dimensional image data as input data referred to by the area information acquisition unit 13 for determining the area of sake rice is the image data acquired by the image processing unit 12 and the image data acquisition unit 11. The process of extracting the 2D liquor rice image data of the area containing only one liquor rice, the process of rotating the 2D liquor rice image so that one liquor rice has a predetermined posture, and adjusting the contrast of the image. It is the processed two-dimensional image data after performing the image processing including the processing. That is, the area information acquisition unit 13 can determine the area of liquor rice from the two-dimensional image data (processed two-dimensional image data) in a state suitable for determining the area of liquor rice, and the determination result is It is more likely to be appropriate.

In step # 13 of FIG. 2, the area information acquisition unit 13 determines the area of sake rice. That is, an area determination step (area information acquisition step) for determining the area of one sake rice in the processed two-dimensional image data generated in the image processing step is executed. In the present embodiment, the area of one sake rice in the processed two-dimensional image data (two-dimensional image data F in FIG. 3) generated by the area information acquisition unit 13 is determined. When there are a plurality of sake rice soaked in water, it is possible to appropriately set how many of the sake rice areas the area information acquisition unit 13 determines. Then, the determination result by the area information acquisition unit 13 is associated with the identifier given to each of the plurality of sake rice and stored in the area storage unit 35b for each sake rice. The area information acquired by the area information acquisition unit 13 is stored in the area storage unit 35b in a state where the change history of the area of one sake rice after the start of immersion in water can be known.

As described above, in the present embodiment, the area information acquisition unit 13 determines the area of one sake rice in the processed two-dimensional image data generated by the image processing unit 12. As a result, a determination result can be obtained in a short time with respect to the area of sake rice.

The area information acquired by the area information acquisition unit 13 can also be output from the output device 3.

Next, the crack rate calculation unit 16 analyzes the two-dimensional image data with reference to the two-dimensional image data acquired by the image data acquisition unit 11, and vertical cracks occur among the plurality of sake rice. Calculate the vertical cracking rate, which is the ratio of sake rice. In addition, "vertical crack" is a state in which the gap between the cracks in the vertical direction generated in sake rice is increased.

In the present embodiment, the crack rate calculation unit 16 includes a morphology determination unit 17 and a morphology data analysis unit 18. For example, the morphological determination unit 17 can automatically determine whether or not cracks have occurred in sake rice using various image recognition techniques. In an example described later, the morphological determination unit 17 will describe a case where the morphology of sake rice is predicted and determined by using an algorithm such as CNN (Convolutional Neural Network). Further, the morphology determination unit 17 will explain a case where machine learning is performed so that information on the morphology of sake rice is output when the processed two-dimensional image data generated by the image processing unit 12 is input. For example, in machine learning, a plurality of weight parameters of CNNs constituting the algorithm of the morphology determination unit 17 are adjusted.

As described above, the two-dimensional image data as input data referred to by the morphological determination unit 17 for determining the morphology of sake rice is 1 in the image data acquired by the image processing unit 12 and the image data acquisition unit 11. A process of extracting two-dimensional liquor rice image data in an area containing only one liquor rice, a process of rotating a two-dimensional liquor rice image so that one liquor rice has a predetermined posture, and a process of adjusting the contrast of the image. It is the processed two-dimensional image data after performing the image processing including. That is, since the morphology determination unit 17 can determine the morphology of sake rice from the two-dimensional image data (processed two-dimensional image data) in a state suitable for determining the morphology of sake rice, the determination result is appropriate. Is more likely to be.

In step # 14 of FIG. 2, the morphological determination unit 17 determines whether or not there is a crack. That is, a form determination step (crack rate calculation step) for determining whether or not one sake rice in the processed two-dimensional image data generated in the image processing step has cracks is executed. In the present embodiment, the morphology determination unit 17 uses the machine learning execution result of the accumulated training data to generate the processed two-dimensional image data (two-dimensional image data F in FIG. 3) generated by the image processing unit 12. It is determined whether or not there is a crack in one sake rice in.

In the present embodiment, the morphological determination unit 17 has a form in which the sake rice is not damaged, a form in which the sake rice is cracked, or a vertical crack in which a vertical cross section is opened along the long axis direction of the sake rice. It is determined whether the rice is in the above-mentioned form or in the form in which the cross-section along the short axis direction of the sake rice is open and the lateral crack is generated. Of these, as shown in Table 1 below, the form in which the sake rice is not damaged and the form in which the sake rice is cracked are judged to be "no cracks", and the form in which the sake rice has vertical cracks. , And, if the sake rice has horizontal cracks, it is judged that there are cracks. When there are a plurality of sake rice soaked in water, it is possible to appropriately set how many of the sake rice morphologies the morphological determination unit 17 determines.

The morphological determination unit 17 has an algorithm constructed so that when the processed two-dimensional image data generated by the image processing unit 12 is input, information regarding the morphology of sake rice is output. For example, the CNN executed by the morphological determination unit 17 is for an input layer into which the processed 2D image data generated by the image processing unit 12 is input and a processed 2D image data input to the input layer. A hierarchical network including a convolutional layer that obtains a feature map by performing convolution processing, a pool layer that shrinks the feature map output from the convolutional layer, a fully connected layer that connects all units, and an output layer. It is composed. The combination of the convolution layer and the pool layer is repeatedly provided a plurality of times, and a plurality of fully connected layers are also provided.

In this embodiment, four types of information (no damage, cracks, vertical cracks, horizontal cracks) are assumed as the information regarding the form of sake rice that is the determination result of the form determination unit 17, and therefore, the unit of the output layer of the CNN. The number will be four. In the output layer, the softmax function can be used as the likelihood function. Since the softmax function divides the value of each unit by the cumulative value of all the units of the output layer and normalizes it, the likelihood of each unit of the output layer takes a value between 0 and 1. Then, the morphology determination unit 17 determines the class having the maximum likelihood as a classification class, that is, the morphology (determination result) of sake rice. Then, the determination result by the morphological determination unit 17 is associated with the identifier given to each of the plurality of sake rice and stored in the morphological storage unit 35d for each sake rice. The morphological storage unit 35d stores the determination result of the morphological determination unit 17 in a state where the change history of the morphology of one sake rice after the start of immersion in water can be known.

FIG. 4 is a diagram showing an example of the state of sake rice. Specifically, FIG. 4A is a diagram showing a form in which sake rice is not damaged, and FIG. 4B is a form in which cracks are generated along the minor axis direction in sake rice. Further, FIGS. 4 (c) and 4 (d) show cracked forms in which the gaps between the cracks generated in the sake rice are large, and the cracks are formed along the short axis direction of the sake rice. There is a form in which a horizontal crack is generated with an open cross section (FIG. 4 (c)) and a form in which a vertical crack with an open vertical cross section along the long axis direction of sake rice is generated (FIG. 4 (d)). As described above, the morphological determination unit 17 has, as the form of cracking, a form in which a horizontal crack having an open cross section along the short axis direction of sake rice or a vertical cross section along the long axis direction of sake rice. It is also determined that is a form in which an open vertical crack has occurred.

As described above, in the present embodiment, the morphology determination unit 17 does not determine the morphology of the sake rice, but the image processing unit 12 generates the image processing unit 12 using the machine learning execution result of the accumulated training data. The morphology of one liquor rice in the processed two-dimensional image data is determined. As a result, it is possible to obtain a consistent determination result in a short time regarding the form of sake rice.

In step # 15 of FIG. 2, the morphological data analysis unit 18 vertically cracks based on the determination result of the morphological determination unit 17 for each of the plurality of sake rice contained in the two-dimensional image data acquired by the image data acquisition unit 11. Calculate the rate. That is, the morphological data analysis step (cracking rate calculation step) for calculating the vertical cracking rate based on the determination result of the morphological determination step is executed. The morphological storage unit 35d stores the determination results of the morphological determination unit 17 for each of the plurality of sake rice in a state where the change history of the morphology of one sake rice after the start of immersion in water can be known. There is. Further, in the morphology storage unit 35d, at least two or more morphologies from the time when the rice is soaked in water to the present (latest) are stored as a change history. Then, the morphological data analysis unit 18 reads out the current (latest) morphology of the plurality of sake rice stored in the morphology storage unit 35d, and among the plurality of sake rice, the sake rice in which vertical cracking occurs. Calculate the vertical crack rate, which is a ratio.

The determination result of the form determination unit 17 can also be output from the output device 3. For example, it is also possible to calculate the ratio (cracking rate) of sake rice in which cracks occur among a plurality of sake rice at any time and output it from the output device 3 in real time. Further, when the determination result of the morphology determination unit 17 is stored in the morphology storage unit 35d in a state where the change history of the morphology of one sake rice after the start of immersion in water can be known, the cracking rate It is also possible to calculate the temporal change of the above and output it from the output device 3.

The estimation unit 19 acquires the vertical cracking rate of sake rice calculated by the cracking rate calculation unit 16, and determines the obtained vertical cracking rate, the predetermined vertical cracking rate of sake rice, and the rate of change in shape in the thickness direction. Based on the correlation of, the rate of change in shape of sake rice in the thickness direction is estimated.

In the process of researching a method for determining the water absorption state such as the water absorption rate and the water absorption distribution of sake rice by observing an image, the inventor of the present application determines the vertical cracking rate, which is the ratio of the sake rice in which vertical cracking occurs. , We obtained a new finding that there is a correlation with the rate of change in shape of sake rice in the thickness direction. Specifically, the inventor of the present application determines the rate of change in the two-dimensional shape of a plurality of sake rice placed in a state of being immersed in water in a plane orthogonal to the thickness direction of the sake rice measured based on two-dimensional image data. By calculating the average value and calculating the average value of the water absorption rates of multiple liquor rice measured using a moisture content meter at the time of measuring the change rate of the two-dimensional shape, the change in the two-dimensional shape of liquor rice The relationship between the rate and the water absorption rate was plotted. In addition, for a plurality of sake rice placed in a state of being immersed in water, the average value of the change rate of the three-dimensional shape measured by using a known method is calculated, and at the time when the change rate of the three-dimensional shape is measured. By calculating the average value of the water absorption rates of a plurality of sake rice measured using a moisture content meter, the relationship between the change rate of the three-dimensional shape of the sake rice and the water absorption rate was plotted (see FIG. 5). As shown in FIG. 5, the determination coefficient of the approximation line regarding the rate of change of the two-dimensional shape and the water absorption rate is 0.30, whereas the determination coefficient of the approximation line regarding the rate of change of the three-dimensional shape and the water absorption rate is 0.30. It was 0.91, and from this, it was found that the rate of change of the two-dimensional shape had a lower correlation with the water absorption rate than the rate of change of the three-dimensional shape.

Then, the inventor of the present application calculates the shape change rate (thickness change rate) in the thickness direction based on the change rate of the two-dimensional shape and the change rate of the three-dimensional shape, and the change rate of the shape in the thickness direction is Then, the horizontal cracking rate and the vertical cracking rate were plotted (see FIG. 6), and the correlation between the horizontal cracking rate and the vertical cracking rate with respect to the rate of change in shape in the thickness direction was investigated. As a result, as shown in FIG. 6, it was found that there is a high correlation between the vertical cracking rate and the rate of change in shape in the thickness direction. Therefore, for a plurality of varieties, a database relating to the correlation between the vertical cracking rate and the rate of change in shape in the thickness direction is created in advance by the same method as described above, and stored in the storage device 35 as appropriate. Based on the correlation, the rate of change in shape in the thickness direction can be estimated from the vertical crack rate.

Therefore, in step # 16 of FIG. 2, the estimation unit 19 is based on the vertical crack rate calculated by the crack rate calculation unit 16 and the correlation between the above-mentioned vertical crack rate and the rate of change in shape in the thickness direction. , Estimate the rate of change in shape of sake rice in the thickness direction. In this way, the rate of change in shape in the thickness direction estimated by the estimation unit 19 corresponds to the rate of change in shape in the thickness direction estimated by the rate of change estimation device 20A. The rate of change in shape in the thickness direction estimated in this way can be stored in the storage device 35 and output from the output device 3. Further, it is also possible to monitor the temporal variation of the shape change rate in the thickness direction and output from the output device 3 when the shape change rate in the thickness direction reaches a predetermined value.

The state calculation unit 30 is based on the rate of change in shape in the thickness direction estimated by the estimation unit 19 (rate of change in shape in the thickness direction estimated by the rate of change estimation device 20A), and the specific gravity, density or water absorption of sake rice. Calculate the rate. In this embodiment, the specific weight of sake rice is calculated.

Specifically, in step # 17 of FIG. 2, the state calculation unit 30 acquires the area of sake rice from the area information acquisition unit 13, and acquires the rate of change in shape of the sake rice from the estimation unit 19 in the thickness direction. Further, the thickness of sake rice stored in the storage device 35 before being immersed in water is acquired, and the volume of sake rice is calculated by multiplying the change rate, area, and thickness of the shape in the acquired thickness direction. The specific weight of sake rice can be calculated by calculating and dividing the calculated volume by the weight of sake rice that is separately measured and appropriately input. The calculated specific gravity of sake rice can be stored in the storage device 35 and output from the output device 3.

FIG. 7 is a flowchart illustrating a machine learning process that is a part of the rate of change estimation method and the state estimation method. In step # 20, the image data acquisition unit 11 acquires two-dimensional image data of sake rice. That is, the image data acquisition step of acquiring the two-dimensional image data of the sake rice placed in the state of being immersed in water is executed. Then, in step # 21, the image processing unit 12 performs image processing to generate processed two-dimensional image data. After that, in step # 22, the morphological determination unit 17 is composed of a combination of the processed two-dimensional image data generated by the image processing unit 12 and information on the morphology of sake rice contained in the processed two-dimensional image data. The training data is stored in the training data storage unit 35c, and machine learning of the training data is executed in step # 23. For example, by labeling one processed two-dimensional image data with correct answer data determined by a human being for one form of sake rice contained in the processed two-dimensional image data. One training data is created. The training data storage unit 35c stores a plurality of training data created in this way.

In the present embodiment, the softmax function is used as the likelihood function in the output layer of the CNN executed by the morphology determination unit 17. Since the softmax function divides the value of each unit by the cumulative value of all the units of the output layer and normalizes it, the likelihood of each unit of the output layer takes a value between 0 and 1. Then, the morphological determination unit 17 uses an error back propagation method to reduce the error between the estimated data (“probability”) and the correct answer data (“1”) generated in the output layer. , The execution result of machine learning is an algorithm that modifies a plurality of weight parameters of the CNN to be executed and outputs information on the morphology of sake rice when the processed two-dimensional image data generated by the image processing unit 12 is input. Get as.

As described above, the machine learning is performed by using the two-dimensional image data (processed two-dimensional image data) in a state suitable for determining the morphology of the liquor rice as training data, and determining the morphology of the liquor rice described above. Is also performed on the two-dimensional image data (processed two-dimensional image data) in a state suitable for determining the morphology of sake rice. As a result, there is a high possibility that the determination result of the morphology determination unit 17 is appropriate.

Next, the effect obtained by performing machine learning using the processed two-dimensional image data as the training data as in the present embodiment will be described. 8 to 10 are graphs showing the relationship between the number of learnings and the correct answer rate. Specifically, FIG. 8 is a graph showing the relationship between the number of learnings and the correct answer rate when machine learning is performed using the training data in which the correct answer data is labeled in the two-dimensional image data B of FIG. FIG. 9 is a graph showing the relationship between the number of learnings and the correct answer rate when machine learning is performed using the two-dimensional image data D of FIG. 3 as training data. FIG. 10 is a graph showing the relationship between the number of learnings and the correct answer rate when machine learning is performed using the two-dimensional image data F of FIG. 3 as training data.

As shown in FIG. 8, when the two-dimensional image data B is used as training data, the correct answer rate is less than 80% even if the number of learnings reaches 20,000. This 80% correct answer rate corresponds to, for example, the correct answer rate when a person who is skilled to some extent determines the morphology of sake rice from two-dimensional image data. On the other hand, when machine learning is performed using the training data in which the correct answer data is labeled on the two-dimensional image data D, the number of learning times is about 12000 and the correct answer rate exceeds 80%. Further, when machine learning is performed using the training data in which the correct answer data is labeled in the two-dimensional image data F, the number of learning times is about 5,000 and the correct answer rate exceeds 80%. In this way, by using the processed two-dimensional image data after image processing, that is, the two-dimensional image data in a state suitable for determining the morphology of sake rice as training data, the number of learnings can be increased. At the very least, we were able to obtain the effect of increasing the correct answer rate.

As described above, according to the rate of change estimation device 20A and the rate of change estimation method, the two-dimensional image data of a plurality of sake rice arranged in a state of being immersed in water is referred to, and among the plurality of sake rice, the vertical direction is obtained. After calculating the vertical cracking rate, which is the ratio of sake rice with cracks, the calculated vertical cracking rate and the correlation between the predetermined vertical cracking rate of sake rice and the rate of change in shape in the thickness direction. Based on the relationship, the rate of change in shape of sake rice in the thickness direction can be estimated. Therefore, it is possible to easily estimate the rate of change in shape of sake rice in the thickness direction without requiring a device having a complicated structure such as a plurality of photographing devices 2. It is naturally possible to estimate the rate of change in shape in the thickness direction of the rice to be cooked, which expands mainly in the direction orthogonal to the vertical direction due to water absorption, using the rate of change estimation device 20A and the rate of change estimation method. Is.

Further, according to the state estimation device 1A and the state estimation method, the specific gravity of sake rice can be calculated based on the estimated rate of change in shape in the thickness direction, and the specific gravity of sake rice can be easily estimated. Can be done. The specific gravity of general rice including sake rice is greatly related to the coarseness and density of starch in the rice, and the more sparsely starched parts such as heart white, the greater the specific gravity. Therefore, by estimating the specific gravity, it is possible to evaluate the state of the white heart (for example, the size). And, for example, sake rice suitable for sake brewing is said to have a large grain and a large heart white, but if the heart white is too large, cracks during rice milling will increase, so the size of the heart white is within a certain range. Since it is considered preferable to use the inside, it is also possible to evaluate the state of whiteness based on the estimated specific gravity and use this evaluation result to select rice suitable for sake brewing.

[Second Embodiment]
The state estimation device 1B (1) of the second embodiment is different from the above-described embodiment in the content of the machine learning process. The state estimation device 1B of the second embodiment will be described below, but the description of the same configuration as that of the above embodiment will be omitted.

FIG. 11 is a diagram showing the configuration of the state estimation device 1B of the second embodiment. As shown in the figure, the state estimation device 1B of the second embodiment further includes a validity determination unit 36 for determining the validity of the form of sake rice determined by the morphology determination unit 17. The validity determination unit 36 is provided in the crack rate calculation unit 16 of the rice analyzer 10B (10). Then, the morphological determination unit 17 combines existing training data with information on the morphology of sake rice determined to be highly valid by the validity determination unit 36 and processed two-dimensional image data for which the morphology has been determined. Perform machine learning in addition to.

Also in this embodiment, the US analyzer 10B is realized by using one or a plurality of computer devices having an information calculation processing function, an information input / output function, an information storage function, and the like. In that case, the function of the image data acquisition unit 11, the function of the image processing unit 12, the function of the area information acquisition unit 13, and the crack rate calculation unit 16 (the function of the morphology determination unit 17, the function of the morphology data analysis unit 18). The function of the above, the function of the estimation unit 19, and the function of the state calculation unit 30 may be installed in the computer device.

The morphological storage unit 35d stores the determination result of the morphological determination unit 17 in a state where the change history of the morphology of one sake rice after the start of immersion in water can be known. For example, in the morphological storage unit 35d, the determination results of the morphological determination unit 17 from the time when the immersion in water is started to the present (latest) are obtained for each of the plurality of sake rice to be morphologically determined. It is stored in a state where the change history of the form can be understood. As a result, it is possible to know what kind of change history one form of sake rice showed.

FIG. 12 is a flowchart illustrating the machine learning process of the second embodiment. This machine learning process is performed in addition to the machine learning process (FIG. 7) described in the first embodiment. More specifically, in step # 30, the validity determination unit 36 determines the validity of the current form of sake rice determined by the morphology determination unit 17. That is, the validity determination step (cracking rate calculation step) for determining the validity of the form of sake rice determined in the form determination step executed by the form determination unit 17 is executed. FIG. 13 is a flowchart illustrating the validity determination process performed by the validity determination unit 36. In step # 40, the validity determination unit 36 determines whether or not the certainty of the current form of sake rice determined by the form determination unit 17 is equal to or higher than a predetermined value. For example, when the maximum likelihood calculated by the softmax function used in the output layer of the morphological determination unit 17 is the above-mentioned certainty, the maximum likelihood calculated by the softmax function is equal to or higher than a predetermined value (for example, 0). If it is (0.8 or more, etc.), the validity determination unit 36 determines that the certainty of the current form of sake rice is equal to or higher than a predetermined value, and proceeds to step # 41.

Next, in step # 41, the validity determination unit 36 determines whether or not the change history of the form of sake rice conforms to a predetermined standard. The predetermined criteria are the following six patterns. Then, if the change history of the form of sake rice conforms to any of the criteria of the six patterns, the validity determination unit 36 shifts to step # 42 and determines that the validity is high.

Specifically, for example, when the form of one sake rice is continuously viewed, the form in which the rice is cracked does not return to the form in which the rice is not damaged, and the rice is cracked. It does not return to the resulting form. That is, if the morphology of sake rice after the start of immersion in water is appropriately determined by the morphological determination unit 17, the change history should conform to a predetermined change criterion. Therefore, in this feature configuration, the validity determination unit 36 has a certainty of the form of sake rice determined by the form determination unit 17 is equal to or higher than a predetermined value, and the form of sake rice after the start of immersion in water is started. When the change history of No. 1 conforms to a predetermined change criterion, it is determined that the form of sake rice determined by the form determination unit 17 is highly valid. As a result, appropriate training data can be accumulated and machine learning can be performed automatically.

On the other hand, the validity determination unit 36 determines that the certainty is not equal to or higher than the predetermined value in the step # 40, and the change history of the liquor rice morphology does not conform to the predetermined criteria in the step # 41. If it is determined, the process proceeds to step # 43 and it is determined that the validity is low.

As described above, in the validity determination unit 36, the certainty of the form of sake rice determined by the form determination unit 17 is equal to or higher than a predetermined value, and the form of sake rice after the start of immersion in water is started. When the change history conforms to a predetermined change criterion, it is determined that the morphology of the sake rice determined by the morphological determination unit 17 is highly valid, and the certainty of the liquor rice morphology determined by the morphological determination unit 17 is high. Is less than a predetermined value, or if the change history of the morphology of sake rice after starting soaking in water does not meet the predetermined change criteria, the sake rice determined by the morphological determination unit 17 It is judged that the validity of the form of is low.

Next, in step # 31, the combination of the information regarding the form of sake rice determined to be highly valid and the processed two-dimensional image data is used as training data. Then, in step # 32, the training data is added to the existing training data to execute machine learning.

In this way, the validity determination unit 36 determines whether the morphology of the sake rice determined by the morphology determination unit 17 is high or low, and the morphology determination unit 17 determines that the validity is high by the validity determination unit 36. Machine learning is executed by adding a combination of information on the morphology of the sake rice and the processed two-dimensional image data for which the morphology has been determined to the existing training data. That is, it is possible to automatically accumulate appropriate training data and perform machine learning.

[Third Embodiment]
The state estimation device 1C (1) of the third embodiment is different from the above embodiment in that the estimation unit 19 estimates the rate of change of the three-dimensional shape. The state estimation device 1C of the third embodiment will be described below, but the description of the same configuration as that of the above embodiment will be omitted.

FIG. 14 is a diagram showing the configuration of the state estimation device 1C of the third embodiment. As shown in the figure, the state estimation device 1C of the third embodiment includes a two-dimensional shape change rate calculation unit (two-dimensional shape change rate calculation means) 40 provided in the rice analyzer 10C. The two-dimensional shape change rate calculation unit 40 analyzes the two-dimensional image data with reference to the two-dimensional image data acquired by the image data acquisition unit 11, and a direction along a plane orthogonal to the thickness direction of sake rice. The rate of change in the two-dimensional shape of sake rice is calculated.

Therefore, in the present embodiment, the US analyzer 10C is realized by using one or a plurality of computer devices having an information arithmetic processing function, an information input / output function, an information storage function, and the like, in which case. The function of the image data acquisition unit 11, the function of the image processing unit 12, the function of the two-dimensional shape change rate calculation unit 40 (the function of the area determination unit 41, the function of the area data analysis unit 42), and the function of the estimation unit 19. Then, a program for realizing the function of the state calculation unit 30 in the computer device may be installed in the computer device.

In the present embodiment, the two-dimensional shape change rate calculation unit 40 includes an area determination unit 41 and an area data analysis unit 42, and the area determination unit 41 is acquired by the image data acquisition unit 11 and is acquired by the image processing unit 12. With reference to the two-dimensional image data processed by, the area of liquor rice in the direction along the plane orthogonal to the thickness direction of liquor rice can be automatically determined by using various image recognition techniques.

As described above, the two-dimensional image data as input data referred to by the area determination unit 41 for determining the area of sake rice is 1 in the image data acquired by the image processing unit 12 and the image data acquisition unit 11. A process of extracting two-dimensional liquor rice image data in an area containing only one liquor rice, a process of rotating a two-dimensional liquor rice image so that one liquor rice has a predetermined posture, and a process of adjusting the contrast of the image. It is the processed two-dimensional image data after performing the image processing including. That is, since the area determination unit 41 can determine the area of liquor rice from the two-dimensional image data (processed two-dimensional image data) in a state suitable for determining the area of liquor rice, the determination result is appropriate. Is more likely to be.

Further, in the present embodiment, the storage device 35 stores in advance the thickness of the sake rice before being immersed in water and the area along the plane orthogonal to the thickness direction.

FIG. 15 shows a step of calculating the change rate of the two-dimensional shape of sake rice (two-dimensional shape change rate calculation step) included in the change rate estimation method and the state estimation method according to the present embodiment, and the change rate of the two-dimensional shape. Is corrected to the rate of change of the three-dimensional shape to estimate the rate of change of the three-dimensional shape (estimation process), and the process of calculating the specific weight of sake rice based on the estimated rate of change of the three-dimensional shape (state calculation). It is a flowchart explaining a process). Since steps # 50 to # 52 in FIG. 15 are the same as steps # 10 to # 12 in FIG. 2, the details thereof will be omitted.

In step # 53 of FIG. 15, the area determination unit 41 determines the area of sake rice. That is, an area determination step (two-dimensional shape change rate calculation step) for determining the area of one sake rice in the processed two-dimensional image data generated in the image processing step is executed. In the present embodiment, the area of one liquor rice in the processed two-dimensional image data (two-dimensional image data F in FIG. 3) generated by the area determination unit 41 is determined. When there are a plurality of sake rice soaked in water, it is possible to appropriately set how many of the sake rice areas the area determination unit 41 determines. Then, the determination result by the area determination unit 41 is associated with the identifier given to each of the plurality of sake rice and stored in the area storage unit 35b for each sake rice. The area storage unit 35b stores the determination result of the area determination unit 41 in a state where the change history of the area of one sake rice after the start of immersion in water can be known.

As described above, in the present embodiment, the area determination unit 41 determines the area of one sake rice in the processed two-dimensional image data generated by the image processing unit 12. As a result, a determination result can be obtained in a short time with respect to the area of sake rice.

In step # 54 of FIG. 15, the area data analysis unit 42 uses the determination results of the area determination unit 41 for each of the plurality of alcoholic rice contained in the two-dimensional image data acquired by the image data acquisition unit 11. Calculate the rate of change of the two-dimensional shape of. That is, the area data analysis step (two-dimensional shape change rate calculation step) for calculating the change rate of the two-dimensional shape based on the determination result of the area determination step is executed. In the area storage unit 35b, the determination result (that is, the area of each sake rice) of the area determination unit 41 for each of the plurality of sake rice is the area of one sake rice after the start of immersion in water. It is stored in a state where the change history can be understood. Further, in the area storage unit 35b, at least two or more areas from the time when the rice is soaked in water to the present (latest) are stored as a change history. Then, the area data analysis unit 42 reads out the area of the plurality of sake rice at the time of starting the immersion in water and the area of the current (latest) plurality of sake rice stored in the area storage unit 35b. Based on these two areas for each sake rice, the rate of change in the two-dimensional shape for each sake rice is calculated.

The determination result of the area determination unit 41 can be output from the output device 3, or the rate of change in the two-dimensional shape of sake rice can be calculated at any time and output from the output device 3 in real time. Further, when the determination result of the area determination unit 41 is stored in the area storage unit 35b in a state where the change history of the area of one liquor rice after the start of immersion in water can be known, the area or It is also possible to calculate the change rate of the two-dimensional shape over time and output it from the output device 3.

The estimation unit 19 according to the present embodiment acquires the change rate of the two-dimensional shape of sake rice calculated by the two-dimensional shape change rate calculation unit 40, and sets the change rate of the two-dimensional shape and the predetermined vertical crack rate. It is different from each of the above-described embodiments in that the change rate of the two-dimensional shape is corrected to obtain the change rate of the three-dimensional shape based on the correlation with the change rate of the shape in the thickness direction.

As described above, the inventor of the present application has confirmed that the rate of change of the two-dimensional shape has a lower correlation with the water absorption rate than the rate of change of the three-dimensional shape, and the lateral cracking rate with respect to the rate of change of the shape in the thickness direction. And the correlation of the vertical crack rate was investigated, and it was confirmed that there was a high correlation between the vertical crack rate and the rate of change in shape in the thickness direction. Then, the inventor of the present application has found that the degree of influence of vertical cracking, which is an index showing the influence of the vertical cracking rate on the rate of change of the three-dimensional shape of sake rice, can be determined based on the above correlation. In the present embodiment, the vertical crack influence degree is a graph showing the correlation between the vertical crack rate and the rate of change in shape in the thickness direction based on a database obtained in advance from measurement results for a plurality of varieties (see FIG. 6). Is created, and the slope of this graph is determined in advance as the degree of influence of vertical cracks.

Therefore, in # 55 of FIG. 15, the estimation unit 19 corrects the change rate of the two-dimensional shape by multiplying the change rate of the two-dimensional shape calculated by the two-dimensional shape change rate calculation unit 40 by the degree of influence of vertical cracking. Then, it is estimated as the rate of change of the three-dimensional shape. In this way, the rate of change of the three-dimensional shape estimated by the estimation unit 19 corresponds to the rate of change of the three-dimensional shape estimated by the rate of change estimation device 20C. The rate of change of the three-dimensional shape estimated in this way can be stored in the storage device 35 and output from the output device 3. Further, it is also possible to monitor the temporal variation of the change rate of the three-dimensional shape and output it from the output device 3 when the change rate of the three-dimensional shape reaches a predetermined value.

The state calculation unit 30 of the present embodiment has a specific gravity and density of sake rice based on the three-dimensional shape change rate (change rate of the three-dimensional shape estimated by the change rate estimation device 20C) obtained by the estimation unit 19. Or calculate the water absorption rate. In this embodiment, the specific weight of sake rice is calculated.

Specifically, in step # 56 of FIG. 15, the state calculation unit 30 acquires the rate of change of the three-dimensional shape of sake rice from the estimation unit 19, and further immerses it in water, which is stored in advance in the storage device 35. Obtain the thickness of sake rice before brewing and the area in the direction along the plane orthogonal to the thickness direction, and multiply the obtained three-dimensional shape change rate and the thickness and area before immersion in water to make sake. The specific weight of sake rice can be calculated by calculating the volume of rice and dividing the calculated volume by the weight of sake rice that is separately measured and appropriately input. The calculated specific gravity of sake rice can be stored in the storage device 35 and output from the output device 3.

As described above, according to the rate of change estimation device 20C and the rate of change estimation method, the change in the two-dimensional shape of the liquor rice is referred to by referring to the two-dimensional image data of a plurality of liquor rice arranged in a state of being immersed in water. After calculating the rate, the degree of influence of vertical cracking, which is determined by the correlation between the predetermined vertical cracking rate of sake rice and the rate of change in shape in the thickness direction, is multiplied by the rate of change in the two-dimensional shape. The rate of change of the two-dimensional shape can be corrected and estimated as the rate of change of the three-dimensional shape. Therefore, for sake rice that expands not only in the direction orthogonal to the vertical direction due to water absorption but also in the vertical direction, a two-dimensional shape can be easily obtained without requiring a device having a complicated configuration such as a plurality of photographing devices 2. The rate of change in the three-dimensional shape of sake rice can be easily estimated by correcting the rate of change in. Using the rate of change estimation device 20A and the rate of change estimation method, the change rate of the two-dimensional shape of the rice to be cooked, which expands mainly in the direction orthogonal to the vertical direction due to water absorption, is corrected to change the three-dimensional shape. Of course, it is also possible to find the rate.

Further, according to the state estimation device 1C and the state estimation method, the specific gravity of sake rice can be calculated based on the estimated rate of change of the three-dimensional shape, and the specific gravity of sake rice can be easily estimated. can. Therefore, as described above, by estimating the specific gravity, it is possible to evaluate the state of the heart white (for example, the size). For example, the state of the heart white is evaluated based on the estimated specific gravity. It is also possible to select rice suitable for sake brewing by using the evaluation result.

[Another Embodiment]
[1] In the first and second embodiments, the rate of change in shape of sake rice in the thickness direction is estimated, and in the third embodiment, the rate of change in three-dimensional shape of rice is estimated. However, the present invention is not limited to this, and for rice other than sake rice, for example, rice to be cooked, the rate of change in shape in the thickness direction and the rate of change in three-dimensional shape can be estimated.

[2] In each of the above embodiments, the state calculation unit 30 calculates the volume of sake rice, measures the calculated volume separately, and divides by the weight of the sake rice appropriately input to obtain the specific weight of the sake rice. Although it is calculated, the density of sake rice may be calculated by dividing the weight of sake rice by the volume.

[3] In the third embodiment, the state calculation unit 30 determines the correlation between the estimated three-dimensional shape change rate and the pre-stored three-dimensional shape of sake rice and the water absorption rate of sake rice. It is also possible to calculate the water content of sake rice based on. That is, for a plurality of sake rice placed in a state of being immersed in water, the average value of the change rate of the three-dimensional shape obtained by a known method in advance was calculated and used to obtain the change rate of the three-dimensional shape. At the time of measuring the rate of change in the dimensional shape, the average value of the water absorption rates of multiple liquor rice measured using a moisture content meter was calculated, and the relationship between the rate of change in the three-dimensional shape of the liquor rice and the water absorption rate was plotted. Keep it. Then, the approximate lines for the plurality of plotted points are stored in the storage device 35 as the correlation between the rate of change in the three-dimensional shape of sake rice and the water absorption rate referred to by the state calculation unit 30. Then, the water absorption rate of sake rice can be calculated based on the estimated change rate of the three-dimensional shape and the above correlation.
In the first and second embodiments, if the area along the plane orthogonal to the thickness direction of the sake rice before being immersed in water is stored in advance, the state calculation unit 30 can store the area in advance. And the area obtained by the area information acquisition unit 13, the change rate of the two-dimensional shape is multiplied by the change rate of the shape in the thickness direction estimated by the estimation unit 19 to calculate the change rate of the three-dimensional shape. By configuring the above, the water absorption rate of sake rice can be calculated in the same manner as described above.

[4] In the above embodiment, the configurations of the state estimation devices 1A, 1B, 1C and the rate of change estimation devices 20A, 20B, 20C have been described with specific examples, but the configurations can be changed as appropriate.

[5] In the first and second embodiments, an example in which the morphological determination unit 17 predicts and determines the morphology of sake rice using a CNN (Convolutional Neural Network) has been described, but the algorithm constituting the morphological determination unit 17 has been described. Is not limited to the CNN described above, and can be configured using various other algorithms such as a support vector machine (SVM), a k-nearest neighbor algorithm, and a discriminant function.
Further, the form determination unit 17 may be configured to determine whether or not the sake rice is cracked by using various image recognition techniques different from those in the above embodiment. Alternatively, the morphological determination unit 17 may be configured to receive a determination result of whether or not the sake rice is cracked, which is performed by a human, and use the result as its own determination result.

[6] The configurations disclosed in the above embodiment (including another embodiment) can be applied in combination with the configurations disclosed in other embodiments as long as there is no contradiction, and the present specification. The embodiments disclosed in the document are examples, and the embodiments of the present invention are not limited thereto, and can be appropriately modified without departing from the object of the present invention.

INDUSTRIAL APPLICABILITY The present invention can be used as a change rate estimation device capable of estimating the rate of change in shape of rice in the thickness direction or the rate of change in three-dimensional shape of rice, and the rate of change in shape and three-dimensional shape in the estimated thickness direction. It can be used as a state estimation device that can estimate the specific gravity, density and water absorption rate of rice based on the rate of change of rice.

1 (1A, 1B, 1C) State estimation device 2 Imaging device 3 Output device 10 (10A, 10B, 10C) US analyzer 11 Image data acquisition unit 13 Area information acquisition unit 16 Crack rate calculation unit 19 Estimating unit 20 (20A, 20A, 20B, 20C) Change rate estimation device 30 State calculation unit 40 Two-dimensional shape change rate calculation unit

Claims (14)

Photographing means for photographing the target rice in a state of being immersed in water,
An image data acquisition means for acquiring two-dimensional image data of the target rice photographed by the photographing means and viewing the target rice from the thickness direction.
With reference to the two-dimensional image data acquired by the image data acquisition means, the two-dimensional image data is analyzed, and among a plurality of target rice, a vertical crack in which a vertical cross section along the long axis direction of the target rice is opened. A crack rate calculation means for calculating the vertical crack rate, which is the ratio of the target rice in which
The vertical cracking rate is obtained from the cracking rate calculating means, and the target is based on the vertical cracking rate and the predetermined correlation between the vertical cracking rate of the target rice and the rate of change in shape in the thickness direction. A rate of change estimation device including an estimation means for estimating the rate of change in the shape of rice in the thickness direction. The rate of change estimation device according to claim 1 and
A state including a state calculating means for acquiring the rate of change in shape in the thickness direction from the estimation means and calculating the specific gravity, density or water absorption rate of the target rice based on the acquired rate of change in shape in the thickness direction. Estimator. Area information acquisition that analyzes the two-dimensional image data with reference to the two-dimensional image data acquired by the image data acquisition means and acquires the area of the target rice in a direction along a plane orthogonal to the thickness direction. Equipped with means,
In the state calculation means, the area is acquired from the area information acquisition means, the rate of change in shape in the acquired thickness direction, the acquired area, and the thickness of the target rice before being immersed in pre-stored water. The state estimation device according to claim 2, wherein the specific gravity or density of the target rice is calculated based on the volume of the target rice calculated based on the above. Photographing means for photographing the target rice in a state of being immersed in water,
An image data acquisition means for acquiring two-dimensional image data of the target rice photographed by the photographing means and viewing the target rice from the thickness direction.
With reference to the two-dimensional image data acquired by the image data acquisition means, the two-dimensional image data is analyzed, and the rate of change of the two-dimensional shape of the target rice in the direction along the plane orthogonal to the thickness direction is determined. Two-dimensional shape change rate calculation means to calculate,
The change rate of the two-dimensional shape is acquired from the two-dimensional shape change rate calculation means, and among a plurality of target rice, the target rice in which a vertical crack having a vertical cross section open along the long axis direction of the target rice is generated. Based on the correlation between the vertical cracking rate, which is the ratio of A rate of change estimation device including an estimation means for estimating the rate of change. A claim for correcting the rate of change of the two-dimensional shape to the rate of change of the three-dimensional shape by multiplying the acquired rate of change of the two-dimensional shape by the degree of influence of vertical cracks determined based on the correlation in the estimation means. Item 4. The rate of change estimation device according to item 4. The rate of change estimation device according to claim 4 or 5,
A state estimation device including a state calculating means for acquiring the change rate of the three-dimensional shape from the estimation means and calculating the specific gravity, density or water absorption rate of the target rice based on the acquired three-dimensional shape change rate. .. In the state calculation means, it is calculated based on the change rate of the acquired three-dimensional shape, the thickness of the target rice before being immersed in water stored in advance, and the area in the direction along the plane orthogonal to the thickness direction. The state estimation device according to claim 6, wherein the specific gravity or density of the target rice is calculated based on the volume of the target rice. The shooting process of shooting the target rice soaked in water,
Based on the two-dimensional image of the target rice taken and viewed from the thickness direction, vertical cracks with an open vertical cross section along the long axis direction of the target rice occur among a plurality of target rice. The crack rate calculation process that calculates the vertical crack rate, which is the ratio of the target rice,
Based on the calculated vertical cracking rate and the predetermined correlation between the vertical cracking rate of the target rice and the shape change rate in the thickness direction, the change rate of the shape of the target rice in the thickness direction is determined. A rate of change estimation method that performs an estimation process and an estimation process. After performing the change rate estimation method according to claim 8, the state calculation for calculating the specific gravity, density or water absorption rate of the target rice based on the change rate of the shape in the thickness direction estimated by the change rate estimation method. A state estimation method that executes a process. The area information acquisition step for acquiring the area of the target rice in the direction along the plane orthogonal to the thickness direction is provided based on the two-dimensional image of the target rice taken and viewed from the thickness direction. ,
In the state calculation step, the target rice is calculated based on the rate of change in shape in the thickness direction, the acquired area, and the volume of the target rice before being immersed in water. The state estimation method according to claim 9, wherein the specific gravity or density of rice is calculated. The shooting process of shooting the target rice soaked in water,
Based on the two-dimensional image of the target rice taken and viewed from the thickness direction, the rate of change in the two-dimensional shape of the target rice in the direction along the plane orthogonal to the thickness direction is calculated. 2 Dimensional shape change rate calculation process and
Correlation between the vertical cracking rate, which is the ratio of the target rice having vertical cracks with an open vertical cross section along the long axis direction of the target rice, and the rate of change in shape in the thickness direction. , And an estimation step of correcting the change rate of the two-dimensional shape and estimating it as the change rate of the three-dimensional shape based on the calculated change rate of the two-dimensional shape. A claim for correcting the rate of change of the two-dimensional shape to the rate of change of the three-dimensional shape by multiplying the calculated rate of change of the two-dimensional shape by the degree of influence of vertical cracks determined based on the correlation in the estimation step. Item 11. The method for estimating the rate of change according to Item 11. A state in which the specific gravity, density, or water absorption rate of the target rice is calculated based on the change rate of the three-dimensional shape estimated by the change rate estimation method after the change rate estimation method according to claim 11 or 12. A state estimation method that executes the calculation process. In the state calculation step, the calculation is made based on the rate of change of the three-dimensional shape and the thickness of the target rice before being immersed in water stored in advance and the area along the plane orthogonal to the thickness direction. The state estimation method according to claim 13, wherein the specific gravity or density of the target rice is calculated based on the volume of the target rice.

Images