Analytical method and analytical apparatus

Abstract

This invention provides an analytical method that can automatically analyze the properties of dough containing grain flour. [Solution] An analytical method for analyzing the properties of dough containing grain flour using an analytical device, comprising: an image acquisition step in which the acquisition unit of the analytical device acquires a group of analytical images consisting of a plurality of analytical images taken of the dough while it is being kneaded in a mixer; and an analytical step in which the analysis unit of the analytical device analyzes the properties by analyzing the plurality of analytical images.

Description

Technical Field

[0001] The present invention relates to an analysis method and an analysis apparatus for analyzing the properties of dough containing flour, particularly wheat flour.

Background Art

[0002] Wheat flour obtained by milling wheat is a processed agricultural product, and its quality varies depending on the harvest year, harvest location, weather conditions, type of fertilizer, etc. Also, the properties of wheat flour vary greatly depending on the variety of wheat. Furthermore, when using wheat flour, multiple types of wheat flour are often blended, and the properties of the wheat flour also change depending on the blending ratio.

[0003] As means for evaluating the surface state (damage) of, for example, steel materials and concrete other than wheat flour, methods and apparatuses for photographing these surfaces and detecting and evaluating damage from the photographing results have been proposed (see, for example, Patent Document 1 or 2).

[0004] On the other hand, as means for evaluating the properties of wheat flour, for example, means for a skilled craftsman to evaluate with the five senses the properties of the dough that change moment by moment, such as the hardness and color of the dough kneaded in a mixer, are often adopted, and the evaluator needs to acquire craftsmanship that cannot be acquired overnight.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0006] As mentioned above, evaluating the properties of fabrics by evaluators requires a great deal of time to acquire the necessary skills, and differences in evaluator skill and individual differences can lead to variations in the evaluation results, making automation desirable.

[0007] The object of the present invention is to provide an analytical method and an analytical apparatus that can automatically analyze the properties of dough containing cereal flour. [Means for solving the problem]

[0008] The present invention relates to an analytical method for analyzing the properties of a dough containing cereal flour using an analytical device, characterized in that the analytical device includes an image acquisition step in which the acquisition unit acquires a group of analytical images consisting of a plurality of analytical images taken of the dough being kneaded in a mixer, and an analysis step in which the analytical unit of the analytical device analyzes the properties by analyzing the plurality of analytical images.

[0009] Furthermore, the analysis method of the present invention is characterized in that the analysis step includes a feature value acquisition step of acquiring the feature value of each pixel constituting the analysis image; an average feature value calculation step of calculating the average feature value of the analysis image from the feature value of each pixel; a standard feature value calculation step of calculating the standard feature value of the analysis image group from the average feature value of each analysis image; a selection step of calculating the difference between the feature value of each pixel and the standard feature value, and selecting the pixels whose difference exceeds a predetermined value as the target pixels; a count calculation step of calculating the number of target pixels for each analysis image; and an average count calculation step of calculating the average number of target pixels in the analysis image group from the number of target pixels in each analysis image, and the properties are analyzed based on the calculation result of the average count calculation step. Furthermore, the analysis method of the present invention is characterized in that the feature value is luminance.

[0010] Furthermore, the selection step of the analysis method of the present invention is characterized in that the pixels whose difference exceeds the predetermined value, and which are connected to a predetermined number or more other pixels whose difference exceeds the predetermined value, are selected as the target pixels.

[0011] Furthermore, the analysis method of the present invention is characterized in that the analysis step includes: an intensity acquisition step of acquiring the edge intensity detected by the Canny edge detection unit of the analysis device for each pixel constituting the analysis image; a selection step of selecting as target pixels pixels whose edge intensity exceeds a first predetermined value, and pixels whose edge intensity is less than or equal to the first predetermined value but greater than or equal to a second predetermined value which is less than the first predetermined value and which are connected to other pixels that exceed the first predetermined value; a number calculation step of calculating the number of target pixels for each analysis image; and an average number calculation step of calculating the average number of target pixels in the group of analysis images from the number of target pixels in each analysis image, and the properties are analyzed based on the calculation result of the average number calculation step.

[0012] Furthermore, the analysis method of the present invention is characterized in that the analysis step includes: an intensity acquisition step of acquiring the edge intensity detected by the Canny edge detection unit of the analysis device for each pixel constituting the analysis image; a selection step of selecting as target pixels pixels whose edge intensity exceeds a first predetermined value, and pixels whose edge intensity is less than or equal to the first predetermined value but greater than or equal to a second predetermined value which is less than the first predetermined value and which are connected to other pixels that exceed the first predetermined value; and a fractal dimension calculation step of calculating the fractal dimension for the edge image created by the target pixels for each analysis image, and the properties are analyzed based on the calculation results of the fractal dimension calculation step.

[0013] Furthermore, the analysis method of the present invention includes a smoothing step of smoothing the analysis image acquired in the image acquisition step, and the analysis step is characterized by analyzing the smoothed analysis image.

[0014] Furthermore, the analysis method of the present invention is characterized in that the analysis step uses an image recognition model that has been trained using a plurality of training images taken of the dough being kneaded in the mixer as input values ​​and the analysis results of the training images as output values ​​to analyze the analysis image.

[0015] Furthermore, the analysis method of the present invention includes a selection step of selecting an image from an image of the dough being kneaded in the mixer in which the kneading member of the mixer is included in a predetermined area or less within the area in which the kneading member of the mixer kneads the dough and in a partial area used for analyzing the properties, and the image acquisition step is characterized by acquiring the analysis image selected in the selection step.

[0016] Furthermore, the analysis method of the present invention includes a elimination step of eliminating a predetermined number of images from images of the dough being kneaded in the mixer, in order of the number of images containing the kneading member within the region where the kneading member of the mixer kneads the dough and in the partial region used for analyzing the properties, and the image acquisition step is characterized in that it acquires images other than those eliminated in the elimination step as the analysis images.

[0017] Furthermore, the analysis method of the present invention is characterized in that, when the analysis step analyzes the analysis image, the region in which the mixing member of the mixer is included within the region in which the mixing member is used for analyzing the properties is excluded from the analysis target.

[0018] Furthermore, the analysis method of the present invention is characterized in that the analysis image is a frame image constituting a moving image, and the group of analysis images is the moving image. Furthermore, the cereal flour in the analysis method of the present invention is characterized in that it contains wheat flour.

[0019] Furthermore, the present invention is an analytical device for analyzing the properties of dough containing cereal flour, and is characterized by comprising: an image acquisition unit that acquires a group of analytical images consisting of a plurality of analytical images taken of the dough while it is being kneaded in a mixer; and an analysis unit that analyzes the properties by analyzing the plurality of analytical images.

[0020] Furthermore, the analysis apparatus of the present invention comprises: a feature value acquisition unit that acquires the feature value of each pixel constituting the analysis image; an average feature value calculation unit that calculates the average feature value of the analysis image from the feature value of each pixel; a standard feature value calculation unit that calculates the standard feature value of the analysis image group from the average feature value of each analysis image; a selection unit that calculates the difference between the feature value of each pixel and the standard feature value and selects the pixels whose difference exceeds a predetermined value as the target pixels; a count calculation unit that calculates the number of target pixels for each analysis image; and an average count calculation unit that calculates the average number of target pixels in the analysis image group from the number of target pixels in each analysis image, wherein the properties are analyzed based on the calculation results by the average count calculation unit. Furthermore, the analysis apparatus of the present invention is characterized in that the feature value is luminance.

[0021] Furthermore, the selection unit of the analytical apparatus of the present invention is characterized in that it selects as the target pixel a pixel whose difference exceeds the predetermined value and which is connected to a predetermined number or more other pixels whose difference exceeds the predetermined value.

[0022] In addition, the analysis apparatus of the present invention includes a Canny edge detection unit that detects an edge intensity for each pixel constituting the analysis image by the analysis unit, pixels for which the edge intensity exceeds a first predetermined value, and pixels for which the edge intensity is less than or equal to the first predetermined value and greater than or equal to a second predetermined value smaller than the first predetermined value and that are continuous with other pixels that exceed the first predetermined value, a selection unit that selects the pixels as target pixels, a count calculation unit that calculates the number of target pixels for each analysis image, and an average count calculation unit that calculates an average number of target pixels in the analysis image group from the number of target pixels in each analysis image, and analyzes the property based on the calculation result by the average count calculation unit.

[0023] In addition, the analysis apparatus of the present invention includes a Canny edge detection unit that detects an edge intensity for each pixel constituting the analysis image by the analysis unit, pixels for which the edge intensity exceeds a first predetermined value, and pixels for which the edge intensity is less than or equal to the first predetermined value and greater than or equal to a second predetermined value smaller than the first predetermined value and that are continuous with other pixels that exceed the first predetermined value, a selection unit that selects the pixels as target pixels, and a fractal dimension calculation unit that calculates a fractal dimension for an edge image created by the target pixels for each analysis image, and analyzes the property based on the calculation result by the fractal dimension calculation unit.

[0024] In addition, the analysis apparatus of the present invention includes a smoothing unit that smooths the analysis image acquired by the image acquisition unit, and the analysis unit analyzes the smoothed analysis image.

[0025] In addition, the analysis apparatus of the present invention includes a storage unit that stores an image recognition model that is machine-learned using a plurality of learning images obtained by imaging the dough being kneaded by the mixer by the analysis unit as input values and the analysis results of the learning images as output values, and analyzes the analysis image using the image recognition model.

[0026] Further, the analyzer of the present invention includes a selection unit that selects, as the analysis image, an image in which the kneading member is included in a partial region used for the analysis of the property within a region where the kneading member of the mixer kneads the dough, and the kneading member is included in a predetermined region or less in the partial region, and the image acquisition unit acquires the analysis image selected by the selection unit.

[0027] Further, the analyzer of the present invention includes an exclusion unit that excludes a predetermined number of images in descending order of the area where the kneading member is included in a partial region used for the analysis of the property within a region where the kneading member of the mixer kneads the dough, and the image acquisition unit acquires, as the analysis image, an image other than the image excluded by the exclusion unit.

[0028] Further, when the analysis unit analyzes the analysis image, the analyzer of the present invention excludes, from the analysis target, a region where the kneading member is included in a partial region used for the analysis of the property within a region where the kneading member of the mixer kneads the dough.

[0029] Further, in the analyzer of the present invention, the analysis image constitutes a frame image of a moving image, and the analysis image group is the moving image. Further, the flour of the analyzer of the present invention contains wheat flour.

Advantages of the Invention

[0030] According to the present invention, it is possible to provide an analysis method and an analyzer capable of automatically analyzing the properties of dough containing flour.

Brief Description of the Drawings

[0031] [Figure 1] It is a block diagram showing a system configuration of an image acquisition device according to the first embodiment. [Figure 2] It is a diagram for explaining a shooting area where a shooting device according to the first embodiment shoots a moving image. [Figure 3] This is a schematic diagram illustrating the frame images constituting the video image according to the first embodiment and the time at which the video image was captured. [Figure 4] This is a diagram illustrating the pixels that constitute an image according to the first embodiment. [Figure 5] This is a block diagram showing the system configuration of the analytical apparatus according to the first embodiment. [Figure 6] This is a schematic diagram illustrating the analysis images that constitute the video image group according to the first embodiment. [Figure 7] This is a flowchart illustrating the analysis method according to the first embodiment. [Figure 8] This is a block diagram showing the system configuration of the analytical apparatus according to the second embodiment. [Figure 9] This is a flowchart illustrating the analysis method according to the second embodiment. [Figure 10] This is a block diagram showing the system configuration of the analytical apparatus according to the third embodiment. [Figure 11] This is a flowchart illustrating the analysis method according to the third embodiment. [Figure 12] This is a block diagram showing the system configuration of the analytical apparatus according to the fourth embodiment. [Figure 13] This is a flowchart illustrating the analysis method according to the fourth embodiment. [Modes for carrying out the invention]

[0032] The following describes an analytical apparatus and analytical method according to a first embodiment of the present invention with reference to the drawings. The analytical apparatus according to this first embodiment is an apparatus for analyzing the properties of wheat flour dough, which is made by adding water to wheat flour and kneading it when making noodles, bread, confectionery, etc. In each embodiment shown below, the case of analyzing the properties of wheat flour dough will be used as an example, but the present invention can also be applied to analyzing the properties of dough that mainly consists of wheat flour or dough that contains grain flour, which is ground flour other than wheat flour. Furthermore, in each embodiment shown below, the case of analyzing the properties of dough kneaded with a vertical mixer will be used as an example, but the present invention can also be applied to analyzing the properties of dough kneaded with a mixer other than a vertical mixer, such as a horizontal mixer.

[0033] The analysis apparatus according to the first embodiment acquires multiple analysis images of the wheat flour dough being mixed in a mixer from an image acquisition device when analyzing the properties of the wheat flour dough. The image acquisition device is a device for acquiring analysis images used in the analysis apparatus to analyze the properties of the wheat flour dough. Figure 1 is a block diagram illustrating the system configuration of the image acquisition device. As shown in Figure 1, the image acquisition device 2 is equipped with a control unit 4 that comprehensively controls each part of the image acquisition device 2. The control unit 4 is connected to an acquisition unit 6, an image processing unit 8, a discrimination unit 10, a storage unit 12, a elimination unit 14, and a transmission unit 15.

[0034] The acquisition unit 6 acquires a video image M (see Figure 3) captured by a shooting device such as a digital camera (not shown). The video image M captures the wheat flour dough (hereinafter simply referred to as "dough") being placed in the bowl 22 of a vertical mixer as shown in Figure 2 and mixed by the hook (kneading member) 24 of the vertical mixer.

[0035] The image processing unit 8 extracts multiple frame images F1 to Fn (n; n is a natural number) necessary for property analysis from all frame images F (see Figure 3) of the moving image M acquired by the acquisition unit 6. In this embodiment, the state of the dough at a predetermined time period T1 within the time T from when water is added to the flour to when it becomes dough in the kneading process is analyzed to evaluate the properties of the dough and any roughness (defects) in the dough. Therefore, the image processing unit 8 extracts n frame images F1 to Fn that constitute the moving image M1 captured at the predetermined time period T1. Furthermore, the image processing unit 8 performs a trimming process to cut out a partial region A (see Figure 2) from the frame images F1 to Fn that show the entire inside of the bowl 22, which is within the area where the kneading member, the hook 24, kneads the dough and is used for analyzing the properties of the dough.

[0036] The discrimination unit 10 determines whether an image can be selected as an analysis image from n images S1 to Sn that have been trimmed by the image processing unit 8, in which the hook 24 is contained in a predetermined area or less within a certain region A. Specifically, the discrimination unit 10 analyzes the values ​​of R (red: integer from 0 to 255), G (green: integer from 0 to 255), and B (blue: integer from 0 to 255) for each pixel in each of the images S1 to Sn.

[0037] The discrimination unit 10 determines whether the analysis results, i.e., the values ​​of R, G, and B (R, G, B), are within a predetermined range. The predetermined range is at least one range of R, G, and B that indicates the color of the imaged hook 24 (R1~R2, G1~G2, B1~B2) (R1, G1, and B1 are integers from 0 to 254, and R2, G2, and B2 are integers from 1 to 255), which is set in advance and stored in the storage unit 12.

[0038] For example, in the image S1 shown in Figure 4, the discrimination unit 10 determines the (R, G, B) values ​​of all pixels P11 to Pxy, and marks pixel P11 only if it determines that the (R, G, B) values ​​of pixel P11 are within a predetermined range (R1 to R2, G1 to G2, B1 to B2). Similarly, the discrimination unit 10 marks pixels P12 to Pxy only if it determines that the (R, G, B) values ​​of pixels P12 to Pxy are within a predetermined range (R1 to R2, G1 to G2, B1 to B2).

[0039] Furthermore, the discrimination unit 10 determines whether pixels that have been determined to be within a predetermined range, i.e., marked pixels, form a region of a predetermined number or more. For example, in a region formed by adjacent pixels being marked consecutively, such as region A1 shown in Figure 4, the discrimination unit 10 maintains the marking of the pixels forming the region if the number of pixels forming this region is greater than or equal to the predetermined number. On the other hand, the discrimination unit 10 erases the marking of pixels where adjacent pixels are not marked, such as pixel P14 shown in Figure 4. Also, in a region A2 shown in Figure 4, where adjacent pixels are marked consecutively, but the number of pixels forming this region is less than the predetermined number, the discrimination unit 10 erases the marking of the pixels forming this region as well. The predetermined number is set in advance and stored in the storage unit 12.

[0040] Furthermore, the discrimination unit 10 calculates the total number of marked pixels in image S1 and determines whether the calculated total number is greater than or equal to a threshold. If it is determined that the total number of marked pixels is greater than or equal to the threshold, the discrimination unit 10 determines that image S1 cannot be selected as an analysis image. On the other hand, if it is determined that the total number of marked pixels is less than the threshold, the discrimination unit 10 determines that image S1 can be selected as an analysis image. The threshold is set in advance and stored in the storage unit 12.

[0041] The discrimination unit 10, similar to image S1, determines the (R, G, B) values ​​of all pixels in each of images S2 to Sn. Furthermore, similar to image S1, the discrimination unit 10 marks pixels in each of images S2 to Sn that it determines have (R, G, B) values ​​within a predetermined range, and erases the markings of pixels that do not form a predetermined number of consecutive regions. Similar to image S1, the discrimination unit 10 also determines whether each of images S2 to Sn can be selected as an analysis image based on the total number of marked pixels.

[0042] As described above, the storage unit 12 stores a predetermined range used for analyzing images S1 to Sn, a predetermined number and threshold used for discrimination by the discrimination unit 10, etc. The exclusion unit 14 excludes images that the discrimination unit 10 has determined not to be selectable as analysis images. The transmission unit 15 transmits a group of images, i.e., a group of analysis images, that the discrimination unit 10 has determined to be selectable as analysis images, to the analysis device 1.

[0043] Figure 5 is a block diagram showing the system configuration of the analysis device 1 according to the first embodiment. As shown in Figure 5, the analysis device 1 includes a control unit 20 that comprehensively controls each part of the analysis device 1. The control unit 20 is connected to an image group acquisition unit 22, an analysis unit 24, and a storage unit 38.

[0044] The image acquisition unit 22 acquires an analysis image group C from the image acquisition device 2, which consists of multiple analysis images U1 to Un taken of wheat flour dough being kneaded in a mixer as shown in Figure 6. The analysis unit 24 includes a brightness acquisition unit 26, an average brightness calculation unit 28, a standard brightness calculation unit 30, a selection unit 32, a count calculation unit 34, and an average count calculation unit 36, and analyzes the multiple analysis images U1 to Un that make up the analysis image group C.

[0045] The luminance acquisition unit 26 acquires the luminance of each pixel that makes up the analysis images U1 to Un. Specifically, the luminance acquisition unit 26 acquires the R, G, and B values ​​for each of the pixels D11 to Dxy (see Figure 6) that make up the analysis image U1, and converts them to luminance using the conversion formula (luminance = 0.229R + 0.587G + 0.114B) to convert the analysis image U1 to grayscale and acquires the luminance Y11 to Yxy for each pixel D11 to Dxy. The luminance acquisition unit 26 also converts the analysis images U2 to Un to grayscale in the same way as the analysis image U1 and acquires the luminance of each pixel. By converting the R, G, and B values ​​to luminance and converting to grayscale, it is possible to reliably detect tears and roughness in the dough even if the color of the dough differs depending on the type of flour and the proportion ratio.

[0046] The average brightness calculation unit 28 calculates the average brightness of the analyzed images U1 to Un from the brightness of each pixel. Specifically, the average brightness calculation unit 28 calculates the average brightness Ya1 of the analyzed image U1 by dividing the sum of the brightness Y11 to Yxy of pixels D11 to Dxy of the analyzed image U1 by the number of pixels in the analyzed image U1. The average brightness calculation unit 28 also calculates the average brightness Ya2 to Yan for the analyzed images U2 to Un in the same manner as for the analyzed image U1.

[0047] The standard brightness calculation unit 30 calculates the standard brightness Ys of the analysis image group C from the average brightness Ya1 to Yan of each of the analysis images U1 to Un. Specifically, the standard brightness calculation unit 30 calculates the standard brightness Ys, which is the value obtained by dividing the sum of the average brightness Ya1 to Yan by the number of analysis images U1 to Un.

[0048] The selection unit 32 calculates the difference V11 to Vxy between the brightness Y11 to Yxy of each pixel D11 to Dxy and the standard brightness Ys, and selects (marks) as target pixels any pixels whose calculated difference exceeds a predetermined value and which are connected to a predetermined number or more other pixels whose calculated difference exceeds the predetermined value. The predetermined value and predetermined number are set in advance and stored in the storage unit 38. Specifically, for example, the selection unit 32 calculates the difference V11 (=Y11-Ys) between the brightness Y11 and the standard brightness Ys of pixel D11 in the analyzed image U1. If the difference V11 does not exceed the predetermined value, pixel D11 is not marked as a target pixel. On the other hand, if the difference V11 exceeds the predetermined value, and pixel D11 is connected to a predetermined number or more other pixels (D12, D21, D22) whose difference V11 exceeds the predetermined value, as shown in region A3 in Figure 6, then pixel D11 is marked as a target pixel as shown in Figure 6. In contrast, even if the difference V11 exceeds a predetermined value, the selection unit 32 does not mark pixel D11 as a target pixel if it is not connected to a predetermined number of other pixels (D12, D21, D22, etc.) whose difference V11 exceeds the predetermined value.

[0049] The selection unit 32 calculates the difference V12 to Vxy between the brightness Y12 to Yxy and the standard brightness Ys for pixels D12 to Dxy, similar to pixel D11. It then determines whether or not to select pixels D12 to Dxy as target pixels based on whether or not the difference exceeds a predetermined value, and if so, whether or not a predetermined number of pixels are connected to other pixels that also exceed the predetermined value.

[0050] Furthermore, the selection unit 32 calculates the difference between each luminance and the standard luminance Ys for each of the analysis images U2 to Un, similar to the analysis image U1. Based on whether the calculated difference exceeds a predetermined value, and if the calculated difference exceeds a predetermined value, whether it is connected to other pixels that exceed the predetermined value in a predetermined number of consecutive pixels or more, the selection unit 32 determines whether to select each pixel constituting each of the analysis images U2 to Un as a target pixel.

[0051] The counting unit 34 calculates the number of target pixels for each of the analysis images U1 to Un. Specifically, the counting unit 34 calculates, for example, the number of pixels selected as target pixels from pixels D11 to Dxy in analysis image U1. Similarly, for analysis images U2 to Un, the counting unit 34 calculates the number of pixels selected as target pixels from the pixels that make up each of the analysis images U2 to Un.

[0052] The average count calculation unit 36 ​​calculates the average number of target pixels in the analysis image group C from the number of target pixels in each of the analysis images U1 to Un. Specifically, the average count calculation unit 36 ​​calculates the average number of target pixels in the analysis image group C, which is the value obtained by dividing the sum of the number of target pixels in each of the analysis images U1 to Un by the number of analysis images U1 to Un.

[0053] The analysis unit 24 analyzes the properties of the fabric based on the calculation result by the average number calculation unit 36, that is, the average number of target pixels in the analysis image group C. Specifically, the analysis unit 24 assigns an evaluation score to assess the properties of the fabric based on the average number of target pixels in the analysis image group C. When the average number of target pixels in the analysis image group is small, the properties of the fabric are good and the evaluation score is high, and when the average number of target pixels in the analysis image group is large, the properties of the fabric are poor and the evaluation score is low.

[0054] The storage unit 38 stores the predetermined values ​​and predetermined numbers used when selecting target pixels in the selection unit 32.

[0055] Next, an analysis method for analyzing the properties of the fabric using the analysis device 1 according to this first embodiment will be described. Figure 7 is a flowchart illustrating the process performed by the control unit 20 to analyze the properties of the fabric.

[0056] First, the control unit 20 instructs the image acquisition unit 22 to acquire an analysis image group C (see Figure 6), which consists of multiple analysis images U1 to Un taken of the dough being kneaded in the mixer, from the image acquisition device 2 (step S10).

[0057] Next, the control unit 20 instructs the brightness acquisition unit 26 to acquire the brightness of pixels D11 that make up the analysis image U1 (step S11). The brightness acquisition unit 26 converts the R, G, and B values ​​of pixels D11 that make up the analysis image U1 into brightness Y11 using a conversion formula. The control unit 20 determines whether or not brightness has been acquired for all pixels D11 to Dxy of the analysis image U1 (step S12). If it has not been acquired (step S12: No), it returns to the process in step S11 and acquires the brightness Y12 for the next pixel D12... The control unit 20 repeats the processes in steps S11 and S12 until it has finished acquiring the brightness Y11 to Yxy for all pixels D11 to Dxy of the analysis image U1 (step S12: No).

[0058] Once the brightness Y11 to Yxy of all pixels D11 to Dxy in the analysis image U1 has been acquired (Step S12: Yes), the control unit 20 instructs the average brightness calculation unit 28 to calculate the average brightness Ya1 of the analysis image U1 (the sum of brightness Y11 to Yxy divided by the number of pixels D11 to Dxy) from the brightness Y11 to Yxy of each pixel D11 to Dxy (Step S13). The control unit 20 determines whether or not the average brightness has been calculated for all analysis images U1 to Un (Step S14). If it has not been calculated (Step S14: No), it returns to the process in Step S11 and acquires the average brightness Ya2 to Un for the next analysis image U2. The control unit 20 repeats the process in Steps S11 to S14 until the average brightness Ya1 to Un for all analysis images U1 to Un has been acquired (Step S14: No).

[0059] Once the average brightness Ya1 to Yan for all analysis images U1 to Un has been obtained (Step S14: Yes), the control unit 20 instructs the standard brightness calculation unit 30 to calculate the standard brightness Ys for the analysis image group C, which is the sum of the average brightness Ya1 to Yan for each analysis image U1 to Un divided by the number of analysis images U1 to Un (Step S15). Next, the control unit 20 instructs the selection unit 32 to calculate the difference V11 between the brightness Y11 of pixel D11 and the standard brightness Ys (Step S16). The control unit 20 determines whether or not the difference V11 to Vxy between the brightness Y11 to Yxy and the standard brightness Ys has been calculated for all pixels D11 to Dxy of the analysis image U1 (Step S17). If it has not been calculated (Step S17: No), it returns to the process in Step S16 and obtains the difference V12 for the next pixel D12... The control unit 20 repeats the processes in steps S16 and S17 until it has finished calculating the difference V11 to Vxy for all pixels D11 to Dxy of the analyzed image U1 (step S17: No).

[0060] Once the difference V11 to Vxy has been calculated for all pixels D11 to Dxy of the analyzed image U1 (Step S17: Yes), the control unit 20 instructs the selection unit 32 to determine whether the difference V11 calculated in Step S16 exceeds a predetermined value (Step S18). If the difference V11 exceeds a predetermined value (Step S18: Yes), the control unit 20 instructs the selection unit 32 to determine whether pixel D11 is connected to other pixels (D12, D21, D22, etc.) whose differences (V12, V21, V22, etc.) exceed a predetermined value in a predetermined number of steps (Step S19). If pixel D11 is connected to other pixels whose differences exceed a predetermined value in a predetermined number of steps (Step S19: Yes), the control unit 20 instructs the selection unit 32 to select (mark) pixel D11 as the target pixel (Step S20).

[0061] On the other hand, if the difference V11 does not exceed a predetermined value (step S18: No), or if pixel D11 is not connected to other pixels whose difference exceeds a predetermined value in a predetermined number of consecutive pixels (step S19: No), the control unit 20 proceeds to the process in step S21.

[0062] The control unit 20 determines whether the processing in steps S18 to S20 has been completed for all pixels D11 to Dxy of the analyzed image U1 (step S21). If it has not been completed (step S21: No), it returns to the processing in step S18 and executes the processing in steps S18 to S21 for pixel D12. The control unit 20 repeats the processing in steps S18 to S21 for pixels D11 to Dxy in order until the processing in steps S18 to S21 has been completed for all pixels D11 to Dxy of the analyzed image U1 (step S21: No).

[0063] When the processing in steps S18 to S21 is completed for all pixels D11 to Dxy of the analysis image U1 (step S21: Yes), the control unit 20 instructs the counting unit 34 to calculate the number of target pixels of the analysis image U1 that were marked in step S20 (step S22). Next, the control unit 20 determines whether or not the number of target pixels has been calculated for all analysis images U1 to Un (step S23). If it has not been calculated (step S23: No), it returns to the process in step S16 and executes the processing in steps S16 to S23 for the analysis image U2. The control unit 20 repeats the processing in steps S16 to S23 for each analysis image U1 to Un in order until the number of target pixels has been calculated for all analysis images U1 to Un (step S23: No).

[0064] Once the number of target pixels has been calculated for all analysis images U1 to Un (Step S23: Yes), the control unit 20 instructs the average number calculation unit 36 ​​to calculate the average number of target pixels in the analysis image group C (the sum of the number of target pixels in each analysis image U1 to Un divided by the number of analysis images U1 to Un) from the number of target pixels in each analysis image U1 to Un calculated in Step S22 (Step S24). Then, the control unit 20 instructs the analysis unit 24 to analyze the properties of the fabric based on the average number of target pixels in the analysis image group C calculated in Step S24 (Step S25).

[0065] According to the analysis apparatus and analysis method of the first embodiment, since the fabric properties are analyzed based on the brightness of each pixel in the analysis images U1 to Un (images in which the hook 24 is not visible), the fabric properties can be automatically and accurately analyzed even for various fabrics (fabrics with different color tones), and tears and roughness in the fabric can be detected. In the first embodiment, luminance was used as an example to explain the feature value of pixels D11 to Dxy, but other quantifiable values ​​such as brightness, saturation, or a specific color (for example, at least one of the R, G, and B values) may also be used as the feature value of pixels D11 to Dxy.

[0066] Next, an analysis apparatus and analysis method according to a second embodiment of the present invention will be described. Note that, for the analysis apparatus according to this second embodiment, the same reference numerals are used for components identical to those shown in Figure 1, and their descriptions are omitted. Figure 8 is a block diagram showing the system configuration of the analysis apparatus according to the second embodiment. As shown in Figure 8, the analysis apparatus 41 includes a control unit 50 that comprehensively controls each part of the analysis apparatus 41. An image group acquisition unit 22, an analysis unit 54, and a storage unit 58 are connected to the control unit 50.

[0067] The analysis unit 54 includes a Canny edge detection unit 55, a selection unit 62, a count calculation unit 34, and an average count calculation unit 36, and analyzes multiple analysis images U1 to Un (see Figure 6) that constitute the analysis image group C.

[0068] The Canny edge detection unit 55 detects edge intensity using the Canny edge detection algorithm to detect edges (contours and boundaries) within the image (analysis images U1 to Un). Specifically, the Canny edge detection unit 55 converts the analysis images U1 to Un to grayscale and detects edge intensity for each pixel constituting the grayscale-converted analysis images U1 to Un (for example, edge intensity E11 to Exy for pixels D11 to Dxy (see Figure 6) constituting the analysis image U1).

[0069] The selection unit 62 selects as target pixels pixels whose edge strength exceeds a first predetermined value Lmax, and pixels whose edge strength is less than or equal to the first predetermined value Lmax but less than the first predetermined value Lmax and greater than or equal to a second predetermined value Lmin (Lmax > Lmin), and which are adjacent to other pixels whose edge strength exceeds the first predetermined value Lmax. That is, (1) if the value of the edge strength exceeds the first predetermined value Lmax, the pixel with that edge strength is selected as a target pixel. Also, (2) if the value of the edge strength is less than or equal to the first predetermined value Lmax and greater than or equal to the second predetermined value Lmin, and which are adjacent to other pixels with an edge strength exceeding the first predetermined value Lmax, then the pixel with that edge strength is selected as a target pixel. On the other hand, (3) if the value of the edge strength is less than or equal to the first predetermined value Lmax and greater than or equal to the second predetermined value Lmin, then which are not adjacent to other pixels with an edge strength exceeding the first predetermined value Lmax, then the pixel with that edge strength is not selected as a target pixel. Also, (4) if the value of the edge strength is less than the second predetermined value Lmin, then the pixel with that edge strength is not selected as a target pixel.

[0070] The analysis unit 54 analyzes the properties of the fabric based on the calculation results from the average number calculation unit 36, i.e., the average number of target pixels in the analysis image group C. The storage unit 58 stores the first predetermined value and the second predetermined value, etc., used when the selection unit 62 selects target pixels.

[0071] Next, an analysis method for analyzing the properties of the dough using the analysis device 41 according to this second embodiment will be described. Figure 9 is a flowchart illustrating the process performed by the control unit 50 to analyze the properties of the dough.

[0072] First, the control unit 50 instructs the image acquisition unit 22 to acquire an analysis image group C (see Figure 6), which consists of multiple analysis images U1 to Un taken of the dough being kneaded in the mixer, from the image acquisition device 2 (step S30).

[0073] Next, the control unit 50 converts the analysis images U1 to Un into grayscale and obtains the edge intensity E11 detected by the Canny edge detection unit 55 for each pixel D11 that makes up the grayscale-converted analysis image U1 (step S31). The control unit 50 determines whether or not edge intensity E11 to Exy has been obtained for all pixels D11 to Dxy of the analysis image U1 (step S32). If it has not been obtained (step S32: No), it returns to the process in step S31 and obtains the edge intensity E12... for the next pixel D12... The control unit 50 repeats the processes in steps S31 and S32 until it has finished obtaining the edge intensity E11 to Exy for all pixels D11 to Dxy of the analysis image U1 (step S32: No).

[0074] Once the edge intensities E11 to Exy of all pixels D11 to Dxy of the analyzed image U1 have been acquired (step S32: Yes), the control unit 50 instructs the selection unit 32 to determine whether the edge intensity E11 acquired in step S31 exceeds a first predetermined value Lmax (step S33). If the edge intensity E11 exceeds the first predetermined value Lmax (step S33: Yes), the control unit 50 instructs the selection unit 32 to perform the process in step S36, i.e., to select (mark) pixel D11 as the target pixel (step S36).

[0075] On the other hand, if the edge strength E11 does not exceed the first predetermined value Lmax (step S33: No), the control unit 50 causes the selection unit 32 to determine whether the edge strength E11 is greater than or equal to the second predetermined value Lmin (step S34). If the edge strength E11 is greater than or equal to the second predetermined value Lmin (step S34: Yes), the control unit 50 determines whether the pixel D11 is connected to another pixel that exceeds the first predetermined value Lmax (step S35). If the pixel D11 is connected to another pixel that exceeds the first predetermined value Lmax (step S35: Yes), the control unit 50 causes the selection unit 32 to perform the process in step S36, i.e., to select (mark) the pixel D11 as the target pixel (step S36).

[0076] On the other hand, if the edge intensity E11 is less than or equal to the second predetermined value Lmin (step S34: No), or if pixel D11 is not connected to any other pixels with an edge intensity exceeding the first predetermined value Lmax (step S35: No), the control unit 50 does not select pixel D11 as a target pixel. That is, (1) if the value of edge intensity E11 exceeds the first predetermined value Lmax, and (2) if the value of edge intensity E11 is less than or equal to the first predetermined value Lmax and greater than or equal to the second predetermined value Lmin, and is connected to any other pixels with an edge intensity exceeding the first predetermined value Lmax, then pixel D11 is selected as a target pixel. On the other hand, (3) if the value of edge intensity E11 is less than or equal to the first predetermined value Lmax and greater than or equal to the second predetermined value Lmin, and is not connected to any other pixels with an edge intensity exceeding the first predetermined value Lmax, and (4) if the value of edge intensity is less than the second predetermined value Lmin, then pixel D11 is not selected as a target pixel.

[0077] The control unit 50 determines whether the processing in steps S33 to S36 has been completed for all pixels D11 to Dxy of the analyzed image U1 (step S37). If it has not been completed (step S37: No), it returns to the processing in step S33 and executes the processing in steps S33 to S37 for pixel D12. The control unit 50 repeats the processing in steps S33 to S37 for pixels D11 to Dxy in order until the processing in steps S33 to S37 has been completed for all pixels D11 to Dxy of the analyzed image U1 (step S37: No).

[0078] When the processing in steps S33 to S37 is completed for all pixels D11 to Dxy of the analysis image U1 (step S37: Yes), the control unit 50 instructs the counting unit 34 to calculate the number of target pixels of the analysis image U1 that were marked in step S36 (step S38). Next, the control unit 50 determines whether or not the number of target pixels has been calculated for all analysis images U1 to Un (step S39). If it has not been calculated (step S39: No), it returns to the processing in step S31 and executes the processing in steps S31 to S39 for the analysis image U2. The control unit 50 repeats the processing in steps S31 to S39 for each analysis image U1 to Un in order until the number of target pixels has been calculated for all analysis images U1 to Un (step S39: No).

[0079] Once the number of target pixels has been calculated for all analysis images U1 to Un (step S39: Yes), the control unit 50 proceeds to steps S40 and S41. Note that the processes in steps S40 and S41 are the same as the processes in steps S24 and S25 shown in Figure 7, so their explanation is omitted.

[0080] According to the analysis apparatus and analysis method of the second embodiment, since the properties of the dough are analyzed based on the edge strength obtained by Canny edge detection in each of the analysis images U1 to Un (images in which the hook 24 is not visible), the properties of the wheat flour dough can be automatically and accurately assessed, and cracks and roughness in the dough can be detected.

[0081] Next, an analysis apparatus and analysis method according to a third embodiment of the present invention will be described. Note that, for the analysis apparatus according to this third embodiment, the same reference numerals are used for components identical to those shown in Figure 8, and their descriptions are omitted. Figure 10 is a block diagram showing the system configuration of the analysis apparatus according to the third embodiment. As shown in Figure 10, the analysis apparatus 71 includes a control unit 80 that comprehensively controls each part of the analysis apparatus 71. An image group acquisition unit 22, an analysis unit 84, and a storage unit 58 are connected to the control unit 80.

[0082] The analysis unit 84 includes a Canny edge detection unit 55, a selection unit 62, and a fractal dimension calculation unit 86, and analyzes multiple analysis images U1 to Un (see Figure 6) that constitute the analysis image group C. The fractal dimension calculation unit 86 calculates the fractal dimension, which is a value representing the degree to which a person recognizes an edge (complexity) in an image, for each analysis image U1 to Un, for the edge image (black and white binarized image) created by the target pixel. The fractal dimension calculation unit 86 also calculates the average fractal dimension of the analysis image group C based on the fractal dimensions of each analysis image U1 to Un.

[0083] The analysis unit 84 analyzes the properties of the fabric based on the calculation results from the fractal dimension calculation unit 86. When the fractal dimension value is small, the edge complexity is low, the fabric properties are good, and the evaluation score is high. When the fractal dimension value is large, the edge complexity is high, the fabric properties are poor, and the evaluation score is low.

[0084] Next, an analysis method for analyzing the properties of the fabric using the analysis apparatus 71 according to this third embodiment will be described. Figure 11 is a flowchart illustrating the process performed by the control unit 80 to analyze the properties of the fabric. Note that the processes in steps S50 to S57 shown in Figure 11 are the same as the processes in steps S30 to S37 shown in Figure 9, so their explanation will be omitted.

[0085] When the processing in steps S53 to S57 is completed for all pixels D11 to Dxy of the analysis image U1 (step S57: Yes), the control unit 80 instructs the fractal dimension calculation unit 86 to calculate the fractal dimension for the edge image created by the target pixels in the analysis image U1 (step S58). Next, the control unit 80 determines whether or not the calculation of the fractal dimension for the edge image created by the target pixels in all analysis images U1 to Un has been completed (step S59). If it has not been completed (step S59: No), it returns to the processing in step S51 and executes the processing in steps S51 to S59 for the analysis image U2. The control unit 80 repeats the processing in steps S51 to S59 for each analysis image U1 to Un in order until the calculation of the fractal dimension for the edge image created by the target pixels in all analysis images U1 to Un has been completed (step S59: No).

[0086] Once the fractal dimension has been calculated for the edge images created by the target pixels for all analysis images U1 to Un (Step S59: Yes), the control unit 80 instructs the fractal dimension calculation unit 86 to calculate the average fractal dimension of the analysis image group C based on the fractal dimensions of each analysis image U1 to Un (Step S60). Next, the control unit 80 instructs the analysis unit 84 to analyze the properties of the fabric from the average fractal dimension of the analysis image group C calculated in Step S60 (Step S61).

[0087] According to the analysis apparatus and analysis method of the third embodiment, since the properties of the dough are analyzed based on the edge strength and fractal dimension obtained by Canny edge detection in each of the analysis images U1 to Un (images in which the hook 24 is not visible), the properties of the wheat flour dough can be automatically and accurately assessed, and cracks and roughness in the dough can be detected.

[0088] In addition, in the second and third embodiments described above, a smoothing unit may be further provided to smooth the analysis images U1 to Un acquired from the image acquisition device 2, and the analysis images U1 to Un may be smoothed before Canny edge detection. An example of a smoothing filter used for image smoothing is a Gaussian filter.

[0089] Next, an analytical apparatus and analytical method according to a fourth embodiment of the present invention will be described. Note that, for the analytical apparatus according to this fourth embodiment, the same reference numerals are used for components identical to those shown in Figure 1, and their descriptions are omitted.

[0090] Figure 12 is a block diagram showing the system configuration of an analysis apparatus according to the fourth embodiment. As shown in Figure 12, the analysis apparatus 91 includes a control unit 90 that comprehensively controls each part of the analysis apparatus 91. The control unit 90 is connected to an image group acquisition unit 22, an analysis unit 92, and a storage unit 94.

[0091] The analysis unit 92 uses the image classification model (image recognition model) 96 stored in the memory unit 94 to analyze the analysis images U1 to Un that make up the analysis image group C acquired by the image acquisition unit 22 from the image acquisition device 2. The image classification model 96 is a learning model that was pre-trained using AI (artificial intelligence) to analyze tears and roughness in the dough shown in the analysis images U1 to Un. It is a model that was trained using multiple training images taken of dough being kneaded by the mixer hook 24 as input values ​​and the analysis results of the training images as output values. The image classification model 96 is an image recognition model constructed using image classification, a type of image recognition technology, and is constructed by following the procedure below. First, the model creator classifies the numerous training images taken for machine learning into five categories: (1) a group of training images with no tears or damage to the fabric (1 point), (2) a group of training images with little or no tears or damage to the fabric (2 points), (3) a group of training images with some tears or damage to the fabric (3 points), (4) a group of training images with tears or damage to the fabric (4 points), and (5) a group of training images with many tears or damage to the fabric (5 points). Next, the numerous training images before classification are used as input values, and the classification result after classification, i.e., one of the scores from the above classifications (1) to (5), is used as the output value for machine learning. Finally, the system is configured to output the probability that the input analysis image falls into each of the above classifications (1) to (5) (for example, for analysis image U11, the probability of (1) being 1 point is 1.1%, (2) being 2 points is 8.9%, (3) being 3 points is 80%, (4) being 4 points is 7.8%, (5) being 5 points is 2.2%, etc.) as output values.

[0092] The analysis unit 92 analyzes whether or not tears or roughness in the fabric are visible in each of the analysis images U1 to Un, based on the output values ​​(probability of corresponding to each of the above classifications (1) to (5)) obtained by inputting each of the analysis images U1 to Un into the image classification model 96, and if so, analyzes the tears or roughness in the fabric. Specifically, the analysis unit 92 calculates for each analysis image U1 to Un the score of the category with the highest probability among the probabilities of each of the above categories (1) to (5) (according to the example of the output values ​​of the analysis image U11 above, the probability of (3) 3 points is the highest at 80%, so 3 points), or the sum of the values ​​of each category (1) to (5) obtained by multiplying the score of each category (1) to (5) by the probability of each category (1) to (5) (according to the example of the output values ​​of the analysis image U11 above, 1 point × 1.1% + 2 points × 8.9% + 3 points × 80% + 4 points × 7.8% + 5 points × 2.2% = 3.011 points).

[0093] The analysis unit 92 then determines the score for analysis image group C as the score of the analysis image group C, which is the score with the most analysis images among the calculated scores for each analysis image U1 to Un, or the average value of the sum of the values ​​for each category (1) to (5) obtained by multiplying the score for each category (1) to (5) by the probability for each category (1) to (5) (the score for each analysis image U1 to Un: 3.011 points in the example of analysis image U11 above). The analysis unit 92 then determines the score for analysis image group C as the score of analysis image group C.

[0094] The memory unit 94 stores image classification models 96 and the like, which are used in the analysis unit 92 to output the probability corresponding to each of the classifications (1) to (5) of the analyzed images U1 to Un.

[0095] Next, an analysis method for analyzing the properties of a fabric using the analysis device 91 according to this fourth embodiment will be described. Figure 13 is a flowchart illustrating the process performed by the control unit 90 to acquire an analysis image.

[0096] First, the control unit 90 instructs the image acquisition unit 22 to acquire an analysis image group C (see Figure 6) from the image acquisition device 2, which consists of multiple analysis images U1 to Un taken of dough being kneaded in the mixer (step S70). Next, the control unit 90 reads the image classification model 96 from the storage unit 96 and instructs the analysis unit 92 to acquire the probability that each of the analysis images U1 to Un corresponds to classification (1) to (5) using the image classification model 96 (step S71). Next, the control unit 90 instructs the analysis unit 92 to calculate the score for each of the analysis images U1 to Un based on the probability acquired in step S71 (step S72).

[0097] Specifically, as described above, the analysis unit 92 calculates for each analysis image U1 to Un the score of the classification with the highest probability from among the probabilities of each classification (1) to (5), or the sum of the values ​​(scores) of each classification (1) to (5) obtained by multiplying the score of each classification (1) to (5) by the probability of each classification (1) to (5).

[0098] The control unit 90 determines whether or not scores have been calculated for all analysis images U1 to Un (step S73). If scores have not been calculated (step S73: No), it returns to the process in step S71 and repeats the processes in steps S71 to S73 until the scores for all analysis images U1 to Un have been calculated (step S73: No).

[0099] Once the scores for all analysis images U1 to Un have been calculated (Step S73: Yes), the control unit 90 instructs the analysis unit 92 to calculate the score for analysis image group C, which is the sum of the scores for each of the analysis images U1 to Un calculated in Step S72 divided by the number of analysis images U1 to Un (Step S74). Next, the control unit 90 instructs the analysis unit 92 to analyze the properties of the fabric from the score of analysis image group C calculated in Step S74 (Step S75).

[0100] According to the analysis apparatus and analysis method of the fourth embodiment, since the properties of the dough are analyzed using the image classification model 96 in each of the analysis images U1 to Un (images in which the hook 24 is not visible), the properties of the wheat flour dough can be automatically and accurately assessed, and cracks and roughness in the dough can be detected.

[0101] In the fourth embodiment described above, an image classification model 96 is used as the image recognition model, but an object detection model or a segmentation model may also be used as the image recognition model.

[0102] The object detection model is an image recognition model constructed using object detection, a type of image recognition technology. First, the model creator annotates a large number of training images taken for machine learning by drawing rectangles around areas with tears or damage to the fabric. Then, using the large number of training images before annotation as input, the model outputs images in which the areas with tears or damage to the fabric are drawn with rectangles.

[0103] When an object detection model is used instead of an image classification model, the analysis unit 92 inputs, for example, the analysis image U1 into the object detection model and obtains the object detection result for the analysis image U1. If an image enclosed in a rectangle is output for the analysis image U1, the analysis unit 92 calculates the area (number of pixels) of the region enclosed in the rectangle and defines this area as the "fabric roughness area." The analysis unit 92 also calculates the "fabric roughness area" for analysis images U2 to Un, and defines the sum of the "fabric roughness area" for each of the analysis images U1 to Un divided by the number of analysis images U1 to Un as the "average fabric roughness area" for the analysis image group C. The analysis unit 92 evaluates that a large "average fabric roughness area" indicates many tears and roughness in the fabric, while a small "average fabric roughness area" indicates few tears and roughness in the fabric.

[0104] A segmentation model is an image recognition model constructed using segmentation, a type of image recognition technology. The model creator first annotates a large number of training images taken for machine learning by filling in areas with tears or damage in the fabric with a specific color. Then, using the unannotated training images as input, the model outputs images in which the areas with tears or damage in the fabric are filled in with the specific color as output.

[0105] When a segmentation model is used instead of an image classification model, the analysis unit 92 inputs, for example, the analysis image U1 into the segmentation model and obtains the object detection result for the analysis image U1. If an image filled with a specific color is output for the analysis image U1, the analysis unit 92 calculates the area (number of pixels) of the region filled with the specific color and defines this area as the "texture roughness area." The analysis unit 92 also calculates the "texture roughness area" for analysis images U2 to Un, and defines the sum of the "texture roughness area" for each of the analysis images U1 to Un divided by the number of analysis images U1 to Un as the "average texture roughness area" for the analysis image group C. The analysis unit 92 evaluates that a large "average texture roughness area" indicates many tears and roughness in the fabric, while a small "average texture roughness area" indicates few tears and roughness in the fabric.

[0106] In the embodiments described above, the example given was that the image group acquisition unit 22 acquires a group of analysis images from the image acquisition device 2. However, the analysis device may also be equipped with an image acquisition device, and the analysis device may select images from images of dough being kneaded in the mixer in which the hooks 24 are included in a predetermined area or less within a partial area A (see Figure 2) used for analyzing the properties of the dough, within the area where the hooks 24 knead the dough. These images may be defined as analysis images U1 to Un.

[0107] Furthermore, in the embodiments described above, the example given was that the exclusion unit 14 of the image acquisition device 2 excludes images in which the hook 24 is visible in an area greater than a predetermined area. However, the configuration may also exclude a predetermined number of images in descending order of the area containing the hook 24 (in descending order of the number of marked pixels). Alternatively, the analysis device may be equipped with an image acquisition device, and the analysis device may exclude a predetermined number of images from images of dough being kneaded in a mixer, in descending order of the area containing the hook 24 within the area where the dough is kneaded and within the partial area A used for analyzing the properties of the dough.

[0108] Furthermore, in the embodiments described above, the example given was the case in which the exclusion unit 14 of the image acquisition device 2 excludes images in which the hook 24 is visible in an area greater than a predetermined area. However, instead of excluding images in which the hook 24 is visible (for example, image S1 in Figure 4), the area in which the hook 24 is visible (for example, area A1 in Figure 4) is excluded from the analysis target, and the area in which the hook 24 is not visible (for example, an area other than area A1 in Figure 4) is made the analysis target, thereby making the image in which the hook 24 is visible (for example, image S1 in Figure 4) the analysis image. Alternatively, if the analysis device is equipped with an image acquisition device, and the analysis device analyzes the analysis image, the area in which the hook 24 is included in the partial area A used for analyzing the properties of the dough, which is within the area in which the dough is kneaded, may be excluded from the analysis target.

[0109] Furthermore, although the above embodiments were described using the example of the case in which the image acquisition device 2 acquires moving images, etc., captured by a photographing device (not shown), the image acquisition device 2 may also be configured to include a photographing unit, and the photographing unit may be configured to photograph the dough while it is being mixed.

[0110] Furthermore, although the above embodiments were described using the example of the image acquisition device 2 acquiring moving images, it is also possible to acquire multiple still images captured in continuous shooting (burst shooting) and use multiple still images captured during a predetermined time period T1 instead of frame images F1 to Fn to select the analysis image. The same applies when the analysis device is equipped with an image acquisition device. [Explanation of Symbols]

[0111] 1, 41, 71, 91...Analysis device, 2...Image acquisition device, 4...Control unit, 6...Acquisition unit, 8...Image processing unit, 10...Discrimination unit, 12...Storage unit, 14...Exclusion unit, 15...Transmission unit, 20, 50, 80, 90...Control unit, 22...Image group acquisition unit, 24, 54, 84, 92...Analysis unit, 26...Brightness acquisition unit, 28...Average brightness calculation unit, 30...Brightness standard calculation unit, 32, 62...Selection unit, 34...Count calculation unit, 36...Average count calculation unit, 38, 58, 94...Storage unit, 55...Canny edge detection unit, 86...Fractal dimension calculation unit, 96...Image classification model.

Claims

1. An analytical method for analyzing the properties of dough containing cereal flour using an analytical device, The acquisition unit of the analysis device includes an image acquisition step of acquiring a group of analysis images consisting of multiple analysis images taken of the dough being kneaded in the mixer, The analysis unit of the analysis device performs an analysis step of analyzing the properties by analyzing multiple analysis images, An analytical method characterized by including [a certain component].

2. The aforementioned analysis process is, A feature value acquisition step is to acquire the feature value of each pixel that makes up the aforementioned analyzed image, A step of calculating an average feature value of the analyzed image from the feature values ​​of each of the aforementioned pixels, A standard feature value calculation step, which calculates the standard feature value of the group of analyzed images from the average feature value of each of the analyzed images, A selection step of calculating the difference between the feature value and the standard feature value of each of the aforementioned pixels, and selecting the pixels whose difference exceeds a predetermined value as the target pixels, A counting step for calculating the number of target pixels for each of the analyzed images, An average number calculation step, which calculates the average number of target pixels in the group of analyzed images from the number of target pixels in each of the analyzed images, Includes, The analysis method according to claim 1, characterized in that the properties are analyzed based on the calculation results obtained by the average number calculation step.

3. The analysis method according to claim 2, characterized in that the aforementioned feature value is luminance.

4. The analysis method according to claim 2, characterized in that the selection step selects as the target pixels pixels whose difference exceeds the predetermined value and which are connected to a predetermined number or more other pixels whose difference exceeds the predetermined value.

5. The aforementioned analysis process is, For each pixel constituting the aforementioned analysis image, an intensity acquisition step is performed to acquire the edge intensity detected by the Canny edge detection unit of the analysis device, A selection step of selecting as target pixels pixels whose edge strength exceeds a first predetermined value, and pixels whose edge strength is less than or equal to the first predetermined value but less than or equal to a second predetermined value and is adjacent to other pixels whose edge strength exceeds the first predetermined value, A counting step for calculating the number of target pixels for each of the analyzed images, An average number calculation step, which calculates the average number of target pixels in the group of analyzed images from the number of target pixels in each of the analyzed images, Includes, The analysis method according to claim 1, characterized in that the properties are analyzed based on the calculation results obtained by the average number calculation step.

6. The aforementioned analysis process is, For each pixel constituting the aforementioned analysis image, an intensity acquisition step is performed to acquire the edge intensity detected by the Canny edge detection unit of the analysis device, A selection step of selecting as target pixels pixels whose edge strength exceeds a first predetermined value, and pixels whose edge strength is less than or equal to the first predetermined value but less than or equal to a second predetermined value and is adjacent to other pixels whose edge strength exceeds the first predetermined value, A fractal dimension calculation step for each of the aforementioned analysis images, which calculates the fractal dimension for the edge image created by the target pixels, Includes, The analysis method according to claim 1, characterized in that the properties are analyzed based on the calculation results obtained by the fractal dimension calculation step.

7. The image acquisition step includes a smoothing step for smoothing the analysis image acquired in the image acquisition step, The analysis method according to claim 5 or 6, characterized in that the analysis step involves analyzing the smoothed analysis image.

8. The analysis method according to claim 1, characterized in that the analysis step involves analyzing the analyzed image using an image recognition model that has been trained using a plurality of training images taken of the dough being kneaded in the mixer as input values ​​and the analysis results of the training images as output values.

9. The process includes a selection step of selecting an image from an image of the dough being kneaded in the mixer, in which the kneading member of the mixer is included in a predetermined area or less within the region where the kneading member of the mixer is kneading the dough and in the partial region used for analyzing the properties, as the analysis image. The analysis method according to claim 1, 2, 3, 4, 5, 6, or 8, characterized in that the image acquisition step acquires the analysis images selected in the sorting step.

10. The process includes a elimination step of eliminating a predetermined number of images taken of the dough being kneaded in the mixer, in order of the number of images containing the kneading member in the region where the kneading member of the mixer kneads the dough and in the partial region used for analyzing the properties, in descending order of quantity. The analysis method according to claim 1, 2, 3, 4, 5, 6, or 8, characterized in that the image acquisition step acquires images other than those excluded in the exclusion step as the analysis images.

11. The analysis method according to claim 1, 2, 3, 4, 5, 6, or 8, characterized in that when analyzing the analysis image, the analysis step excludes from the analysis the area in which the mixing member of the mixer is included in the partial area used for analyzing the properties, which is within the area in which the mixing member of the mixer is used for analyzing the properties.

12. The analysis method according to claim 1, 2, 3, 4, 5, 6, or 8, characterized in that the analysis image is a frame image constituting a moving image, and the group of analysis images is the moving image.

13. The analytical method according to claim 1, 2, 3, 4, 5, 6, or 8, characterized in that the aforementioned grain flour contains wheat flour.

14. An analytical device for analyzing the properties of dough containing cereal flour, An image acquisition unit acquires a group of analysis images consisting of multiple analysis images taken of the dough being kneaded in a mixer, An analysis unit that analyzes the properties by analyzing multiple analysis images, An analytical device characterized by being equipped with the following features.

15. The aforementioned analysis unit is A feature value acquisition unit that acquires feature values ​​for each pixel constituting the aforementioned analyzed image, An average feature value calculation unit calculates the average feature value of the analyzed image from the feature values ​​of each of the pixels, A standard feature value calculation unit calculates the standard feature value of the group of analyzed images from the average feature value of each of the analyzed images, A selection unit calculates the difference between the feature value and the standard feature value for each of the aforementioned pixels, and selects the pixels whose difference exceeds a predetermined value as the target pixels. A counting unit calculates the number of target pixels for each of the analyzed images, An average number calculation unit calculates the average number of target pixels in the group of analyzed images from the number of target pixels in each of the analyzed images, Equipped with, The analytical apparatus according to claim 14, characterized in that it analyzes the properties based on the calculation results by the average number calculation unit.

16. The analytical apparatus according to claim 15, characterized in that the aforementioned feature value is luminance.

17. The analysis apparatus according to claim 15, characterized in that the selection unit selects as the target pixels pixels in which the difference exceeds the predetermined value and which are connected to a predetermined number or more other pixels in which the difference exceeds the predetermined value.

18. The aforementioned analysis unit is A Canny edge detection unit detects the edge intensity for each pixel constituting the aforementioned analysis image, A selection unit that selects as target pixels pixels whose edge strength exceeds a first predetermined value, and pixels whose edge strength is less than or equal to the first predetermined value but less than or equal to a second predetermined value and is adjacent to other pixels whose edge strength exceeds the first predetermined value, A counting unit calculates the number of target pixels for each of the analyzed images, An average number calculation unit calculates the average number of target pixels in the group of analyzed images from the number of target pixels in each of the analyzed images, Equipped with, The analytical apparatus according to claim 14, characterized in that it analyzes the properties based on the calculation results by the average number calculation unit.

19. The aforementioned analysis unit is A Canny edge detection unit detects the edge intensity for each pixel constituting the aforementioned analysis image, A selection unit that selects as target pixels pixels whose edge strength exceeds a first predetermined value, and pixels whose edge strength is less than or equal to the first predetermined value but less than or equal to a second predetermined value and is adjacent to other pixels whose edge strength exceeds the first predetermined value, A fractal dimension calculation unit calculates the fractal dimension for the edge image created by the target pixels in each of the aforementioned analysis images. Equipped with, The analytical apparatus according to claim 14, characterized in that it analyzes the properties based on the calculation results by the fractal dimension calculation unit.

20. The image acquisition unit includes a smoothing unit that smooths the analysis image acquired by the image acquisition unit, The analytical apparatus according to claim 18 or 19, characterized in that the analysis unit analyzes the smoothed analytical image.

21. The aforementioned analysis unit is The device includes a storage unit that stores an image recognition model trained using multiple training images taken of the dough being kneaded in the mixer as input values, and the analysis results of the training images as output values. The analytical apparatus according to claim 14, characterized in that it analyzes the analyzed image using the image recognition model.

22. The device includes a selection unit that, from images of the dough being kneaded in the mixer, selects an image as the analysis image in which the kneading member of the mixer is included in a predetermined area or less within the region where the kneading member of the mixer kneads the dough and in a partial region used for analyzing the properties, The analytical apparatus according to claim 14, 15, 16, 17, 18, 19, or 21, characterized in that the image acquisition unit acquires the analytical images selected in the sorting unit.

23. The system includes a filtering unit that filters out a predetermined number of images taken of the dough being kneaded in the mixer, in descending order of the number of images containing the kneading member within the region where the kneading member of the mixer kneads the dough and which is used for analyzing the properties of the dough, in order of the number of images containing the kneading member. The analysis apparatus according to claim 14, 15, 16, 17, 18, 19, or 21, characterized in that the image acquisition unit acquires images other than those excluded by the exclusion unit as the analysis images.

24. The analysis apparatus according to claim 14, 15, 16, 17, 18, 19, or 21, characterized in that when analyzing the analysis image, the analysis unit excludes from the analysis the area in which the mixing member of the mixer is included in the partial area used for analyzing the properties, which is within the area in which the mixing member of the mixer is used for analyzing the properties.

25. The analysis apparatus according to claim 14, 15, 16, 17, 18, 19, or 21, characterized in that the analysis image is a frame image constituting a moving image, and the group of analysis images is the moving image.

26. The analytical apparatus according to claim 14, 15, 16, 17, 18, 19, or 21, characterized in that the aforementioned grain flour contains wheat flour.

Images