JP6617986B2 - Identification apparatus, information processing apparatus, program, and identification method - Google Patents

Description

  The present invention relates to a discrimination device, an information processing device, a program, and a discrimination method.

  Stock as a cut flower has been cultivated actively and is one of the flowers favored by consumers. Stock is a cruciferous flower that blooms from October to March, and is an annual plant that grows from about 20 cm to a height of about 80 cm. This stock has a single bloom and a double bloom, and the double bloom is more voluminous and has a higher commercial value than the single bloom. Therefore, in order to increase profits, it is important to accurately distinguish between single-flowered seedlings and double-flowered seedlings when the stock is in a seedling state (double discrimination).

  However, when performing an eight-fold discrimination by visual observation, it is necessary to identify subtle differences in cotyledons depending on the operator's senses, which requires skill and time. Therefore, research on techniques that can be easily discriminated even by beginners has been conducted. Examples of the “beginner” of the Yae discrimination include a person who has never experienced an Yae discrimination or an operator of an Yae discrimination who has less than 36 months of experience.

  For example, in Non-Patent Document 1, after a germinated seedling is photographed with a camera from vertically above the tray, image processing for extracting stock cotyledons from the seedling image is performed using a general video camera and a notebook computer, A method of performing double discrimination by calculating feature values and optimizing boundary values between double-flowering seedlings and single-flowering seedlings is disclosed.

Sakaki Yuan, Makoto Doi, Nobuaki Ishizuka, “Development of a precise management robot for flower and vegetable seedlings (1st report)-Identification of stock seedlings by image processing”, Journal of the Agricultural Machinery Society 66 (2), p.68-75, 2004

  Here, in the method disclosed in Non-Patent Document 1, a morphological feature amount and a color feature amount are calculated as feature amounts, and the morphological feature amount includes a plurality of items for each cotyledon form (single leaf or bilobal). It is subdivided and calculated. Further, in the above method, in order to determine a boundary value serving as a discrimination reference, it is necessary to calculate an average value and a standard deviation for each feature amount and convert the feature amount into a relative value.

  Accordingly, when the double discrimination is performed by the above method, the number of items to be calculated as the feature amount increases. In addition to the feature amount, a plurality of types of values must be further calculated, and each of these values must be used for the double discrimination. For this reason, there are many parameters to be considered when performing an eight-fold discrimination, and it has not necessarily been said that even a beginner can easily perform an eight-fold discrimination.

  One embodiment of the present invention has been made in view of the above-described problems, and an object thereof is to enable even a beginner to easily perform double discrimination.

  In order to solve the above-described problem, a discrimination device according to one aspect of the present invention is a discrimination device for discriminating double-flowering seedlings from a plurality of stock seedlings, and from the images of the plurality of stock seedlings, For each of the plurality of stock seedlings, an information acquisition unit that acquires at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value; and the plurality of stock seedlings By determining at least one of whether the color characteristic value is included in a predetermined first range and whether the size value is included in a predetermined second range for each A determining unit that determines whether or not a stock seedling is the double-flowered seedling.

  According to the above configuration, it is determined whether at least one of the color characteristic value and the size value is included in a predetermined numerical range, thereby distinguishing double-flowered seedlings from a plurality of stock seedlings. Accordingly, it is sufficient to use two types of parameters at the maximum in the double discrimination, and the number of parameters to be considered can be smaller than in the past. Moreover, since it is only necessary to determine whether or not a specific value (color characteristic value, size value) relating to a cotyledon is included in a predetermined numerical range, even a beginner can easily determine. Therefore, even a beginner can easily perform double discrimination.

  In the identification device according to an aspect of the present invention, the information acquisition unit acquires a color difference of the cotyledon from the color characteristic value information, and the determination unit determines whether the color difference of a specific seedling of the stock is the first. It may be determined whether or not it is included in at least the upper 50% of all the color differences that are the range.

  According to the above configuration, whether or not a specific stock seedling is a double-flowered seedling is determined only by determining whether or not the color difference of a specific stock seedling is included in at least the top 50% of all the color differences. Can be identified. Therefore, it is sufficient to use only one parameter for the octet discrimination, and the novice can perform the octet discrimination more easily.

  In general, if the correct answer rate for double discrimination is 80% or more, it is considered successful. In that respect, according to the above configuration, even if a beginner sets the lower limit value of the first range to at least the upper 50%, the correct answer rate for the eight-fold discrimination becomes about 80% or more. Therefore, even a beginner can perform an eight-fold discrimination with a correct answer rate that is generally considered to be almost successful.

  In the present specification, among the seedlings of stocks that were confirmed to be double-flowering seedlings by the double-identification, the number of stock seedlings that were actually double-flowering seedlings was determined. The value calculated by dividing by the number of stocks and multiplying by 100 is taken as the “correction rate (%) for double discrimination”. In addition, a case where the correct answer rate of the above-mentioned double discrimination is 75% or more is set as “correct answer rate is about 80% or more”. Furthermore, a case where the correct answer rate of the above-mentioned double discrimination is not less than 75% and not more than 84% is set as “correct answer rate is about 80%”.

  In the identification device according to an aspect of the present invention, the information acquisition unit acquires the saturation of the cotyledon from the color characteristic value information, and the determination unit determines that the saturation of the seedling of the specific stock is the It may be determined whether or not it is included in at least the upper 50% of all the saturations that are the first range.

  According to the above configuration, only by determining whether or not the saturation of a specific stock seedling is included in at least the top 50% of all the saturations, the specific stock seedling is a double bloom seedling. You can distinguish whether or not. Therefore, a beginner can more easily perform double discrimination.

  In addition, by setting the lower limit of the first range to at least the upper 50%, even the beginner can achieve a correct answer rate of about 80% or more for double discrimination. Therefore, even a beginner can perform an eight-fold discrimination with a correct answer rate that is generally considered to be almost successful.

  In the identification device according to one aspect of the present invention, the information acquisition unit acquires a hue angle of the cotyledon from the color characteristic value information, and the determination unit determines whether the hue angle of a specific seedling of the stock is the It may be determined whether it is included in at least the lower 60% of all the hue angles that are the first range.

  According to the above configuration, only by determining whether or not the hue angle of a specific stock seedling is included in at least the lower 60% of all hue angles, the specific stock seedling is a double-flowered seedling. You can distinguish whether or not. Therefore, a beginner can more easily perform double discrimination.

  Further, by setting the lower limit value of the first range to at least the lower 60%, the correct answer rate for the double discrimination becomes about 80% or more even for beginners. Therefore, even a beginner can perform an eight-fold discrimination with a correct answer rate that is generally considered to be almost successful.

  In the identification device according to an aspect of the present invention, the information acquisition unit acquires the leaf area of the cotyledon from the size value information, and the determination unit determines that the leaf area of the seedling of the specific stock is the first value. It may be determined whether or not it is included in at least the top 40% of all the leaf areas that are two ranges.

  According to the above configuration, only by determining whether or not the leaf area of a specific stock seedling is included in at least the top 40% of all leaf areas, the specific stock seedling is a double bloom seedling. You can distinguish whether or not. Therefore, a beginner can more easily perform double discrimination.

  In addition, by setting the lower limit value of the second range to at least the upper 40%, even the beginner can achieve a correct answer rate of about 80% or more for the double discrimination. Therefore, even a beginner can perform an eight-fold discrimination with a correct answer rate that is generally considered to be almost successful.

  In the identification device according to an aspect of the present invention, the information acquisition unit acquires a color difference of the cotyledon and a saturation of the cotyledon from the color characteristic value information, and acquires the leaf area of the cotyledon from the size value information. And (i) whether the color difference of the specific stock seedling is included in at least the top 40% of all the color differences in the first range, and (ii) Whether the saturation of the particular seedling of the stock is included in at least the top 40% of all the saturations in the first range, and (iii) the seedling of the particular stock By determining whether or not the leaf area is included in the top 10% of all the leaf areas in the second range, it is determined whether or not a particular seedling of the stock is the double-flowered seedling. You may decide.

  According to the above configuration, for a specific stock seedling, it is possible to determine whether the specific stock seedling is in double blooming by simply determining whether the cotyledon color difference, saturation, and leaf area are within a predetermined numerical range. You can distinguish whether it is a seedling. Therefore, it is sufficient to use three parameters for the eight-fold discrimination, and even a beginner can easily perform the eight-fold discrimination.

  In addition, by setting the lower limit of color difference / saturation as the first range to at least the upper 40%, and setting the lower limit of the leaf area as the second range to the upper 10%, even beginners can correctly answer the double check. The rate is about 80% or more. Therefore, even a beginner can perform an eight-fold discrimination with a correct answer rate that is generally considered to be almost successful.

  An information processing apparatus according to one aspect of the present invention includes a discrimination apparatus according to each aspect of the present invention. According to the above configuration, it is possible to realize an information processing apparatus that can be easily discriminated even by beginners.

  The discrimination method according to an aspect of the present invention is a discrimination method for discriminating double-flowered seedlings from a plurality of stock seedlings, and from each of the plurality of stock seedlings, from the plurality of stock seedling images, At least one of color characteristic value information representing cotyledon color characteristic information and size value information representing the cotyledon size value information, and for each of the plurality of stock seedlings, the information obtaining step At least one of whether the acquired color characteristic value is included in a predetermined first range and whether the size value acquired in the information acquisition step is included in a predetermined second range. Determining whether the particular seedling of the stock is the double-flowering seedling. According to the above configuration, it is possible to realize a discrimination method that enables even a beginner to easily perform double discrimination.

  The discrimination device according to each aspect of the present invention may be realized by a computer. In this case, the discrimination device is operated by a computer by operating the computer as each unit (software element) included in the discrimination device. A program for a control device to be realized and a computer-readable recording medium on which the program is recorded also fall within the scope of the present invention.

  According to one embodiment of the present invention, even a beginner can easily perform double discrimination using image analysis.

It is a block diagram which shows the functional structure of the imaging device and information processing apparatus which concern on Embodiments 1-3 of this invention. It is a flowchart which shows the discrimination method which concerns on Embodiment 1 of this invention. It is a flowchart which shows the modification of the discrimination method which concerns on Embodiment 1 of this invention. It is a flowchart which shows the modification of the discrimination method which concerns on Embodiment 1 of this invention. It is a flowchart which shows the discrimination method which concerns on Embodiment 2 of this invention. It is a flowchart which shows the discrimination method which concerns on Embodiment 3 of this invention. It is a figure which shows the TIF image which concerns on several stock seedlings. It is a figure which shows the binarized image which binarized the TIF image. (A) is a diagram showing an L ^* image, (b) is a diagram showing an a ^* image, and (c) is an image of b ^* , with respect to the image after the TIF image is converted into the CIELAB color space. FIG. (A) is a figure which shows the calculation result of L value, and the image of L ^* , (b) is a figure which shows the calculation result of a value, and the image of a ^* , (c) is a figure of b value It is a figure which shows a calculation result and the image of b ^* . It is a histogram which shows the calculation result of the color difference of a cotyledon, saturation, leaf area, and hue angle. (A) And (b) is a figure which shows the discrimination result which concerns on Example 1 of this invention. (A) is a figure which shows the image by which the discrimination result based on Example 1 of this invention was displayed. (B) is a figure which shows the image by which the correspondence of the said discrimination result and an actual double bloom seedling was displayed. It is a graph which shows the discrimination result which concerns on Example 1 of this invention. (A) And (b) is a figure which shows the discrimination result which concerns on Example 2 of this invention. It is a graph which shows the discrimination result which concerns on Example 2 of this invention. (A) And (b) is a figure which shows the discrimination result which concerns on Example 3 of this invention. It is a graph which shows the discrimination result which concerns on Example 3 of this invention. (A) is a figure which shows the image by which the discrimination result based on Example 4 of this invention was displayed. (B) is a figure which shows the image by which the correspondence of the said discrimination result and an actual double bloom seedling was displayed. It is a table | surface which shows the discrimination result which concerns on the comparative example of this invention.

Embodiment 1
For convenience of explanation, components having the same functions as those described for specific items are denoted by the same reference numerals, and the description thereof is omitted. The same applies to the following embodiments, examples, and comparative examples.

<Functional configuration of information processing apparatus>
Examples of the information processing apparatus 100 include a PC, a tablet terminal, and a seedling selection robot. The same applies to information processing apparatuses 200 and 300 described later. As illustrated in FIG. 1, the information processing apparatus 100 includes a display unit 1, an operation input unit 2, a storage unit 3, and a control unit 4.

  The display unit 1 displays a discrimination result by the discrimination device 10 described later, an image resulting from execution of various functions (application software) equipped in the information processing device 100, and the like. The operation input unit 2 acquires an input user operation. For example, when the information processing apparatus 100 is a tablet terminal, a touch panel integrated with the display unit 1 is used.

  The storage unit 3 stores various data generated by the discrimination device 10 described later, various control programs executed by the control unit 4, and is configured by a non-volatile storage device such as a hard disk or a flash memory. Is done. The storage unit 3 may be provided outside the information processing apparatus 100, for example.

  The control unit 4 comprehensively controls the information processing apparatus 100 and includes a discrimination device 10. The discrimination device 10 is a device for discriminating double-flowered seedlings from a plurality of stock seedlings (see FIG. 7 and the like), and includes an image analysis unit 11 (information acquisition unit) and a determination unit 12 as shown in FIG. Yes. In FIG. 1, the control unit 4 is built in the information processing apparatus 100, but is not limited to this case. For example, the control unit 4 may be an external device attached to the information processing apparatus 100 or a network server used via a communication unit (not shown). Alternatively, the identification device 10 may be an external device attached to the information processing device 100.

  In each of the following embodiments, after seeding stock seeds one by one in the cell tray, seedlings in a state where the cotyledons have been developed are used as stock seeds (hereinafter referred to as the target of the eight-fold discrimination by the discrimination device 10). Abbreviated as “target seedling”). For example, a seedling in a state where the cotyledon has developed and the main leaf has started to appear (in summer, about 10 days after sowing) may be the target seedling, or a seedling in a state after the main leaf has appeared considerably. (If it is summer, about 16 days after sowing) may be the target seedling. In general, seedlings in a state where cotyledons have developed and real leaves have started to appear are most suitable for double check.

  Here, the cultivation form of stock is not limited to the case of the above-mentioned cell tray. Stock seedlings may be sown, for example, in pots (see [Example] described later) or directly in the field.

<Functional configuration of identification device>
(Image Analysis Department)
The image analysis unit 11 analyzes images of a plurality of target seedlings (stock seedlings) grown on the cell tray as pre-processing for the double discrimination. Specifically, first, a plurality of target seedlings to be subjected to image analysis are imaged from above with the same visual field using an imaging device 500 such as a digital camera. Although there is no restriction | limiting in particular about an imaging location, It is preferable to image in a dark room. Captured images (hereinafter abbreviated as “initial images”) of a plurality of target seedlings are developed in the TIF format by the image editing software of the imaging apparatus 500 from the state of the RAW image, and transmitted to the image analysis unit 11.

Receiving the TIF format initial image (hereinafter abbreviated as “TIF image”), the image analysis unit 11 performs binarization processing on the TIF image in the RGB color space to generate a binarized image. During the binarization processing, the image analysis unit 11 calculates and acquires values (hereinafter, abbreviated as “threshold”) indicating the hue, saturation, and brightness of the boundary between the target seedling and the background in the image. In some cases, the leaf area of the cotyledon may be calculated and acquired during the binarization process (see <Modification> described later). Then, specify a TIF images ^L * is converted into CIELAB color space, after dividing into ^{a *} and ^{b *} each image, the from the binarized image ^L *, in order to ^{a *} and ^{b *} each image To do.

  In each of the following embodiments, the target seedling and the background in the image are segmented by the threshold values of three indices (hue, saturation, and lightness). However, the present invention is not limited to this case. For example, segmentation may be performed using only one index, or segmentation may be performed using an index other than “hue, saturation, and brightness”.

  The CIELAB color space is an almost complete color space established by the International Commission on Illumination (CIE). It can represent all colors that can be seen by the human eye and can be used as a reference for device-specific models for image processing.

The CIELAB color space is composed of three coordinates L ^* , a ^* and b ^* . L ^* represents the lightness of the color, and L ^* = 0 is black and L ^* = 100 is white. The white reflection color has a higher value than L ^* = 100. a ^* represents the position between red / magenta and green, with negative values closer to green and positive values closer to red / magenta. b ^* represents a position between yellow and blue, with negative values closer to blue and positive values closer to yellow.

Next, for each of a plurality of target seedlings, three values (color characteristic value information) of the cotyledon L ^* value, a ^* value, and b ^* value are calculated and acquired. In the following embodiments, examples, and comparative examples, the three values are average values of all the pixels in the L ^* , a ^*, and b ^* images.

Next, the image analysis unit 11 calculates and acquires a cotyledon color difference ΔE (color characteristic value) based on the three values of the L ^* value, the a ^* value, and the b ^* value for each of the plurality of target seedlings. . The color difference represents a distance in the color space. The cotyledon color difference ΔE is calculated using Equation 1 below.
(Formula 1)

Here, in order to give a ranking to all the target seedlings collectively, the color difference ΔE of the cotyledon is based on (L ^* , a ^* , b ^* ) = (0, 0, 0) in the above formula 1. Value. The image analysis unit 11 transmits the calculated / acquired cotyledon color difference ΔE to the determination unit 12. The cotyledon color difference ΔE may be transmitted to the storage unit 3 and stored therein. The same applies to the cotyledon saturation C ^* , the cotyledon leaf area s, and the cotyledon hue angle h described later.

(Decision part)
Based on the cotyledon color difference ΔE received from the image analysis unit 11, the determination unit 12 determines whether or not each of the plurality of target seedlings is a double blooming seedling. Specifically, the determination unit 12 determines whether or not the cotyledon color difference ΔE related to the specific target seedling is included in the upper x% (first range) of the cotyledon color differences ΔE related to all the target seedlings. Determine.

  When it is included in the upper x%, the determination unit 12 determines that the specific target seedling that is the determination target is an double bloom seedling. On the other hand, when it is not included in the x%, the determining unit 12 determines that the specific target seedling is not an eight-flowering seedling, in other words, a single-flowering seedling.

  Here, a numerical value of at least 50 or more is used as the value of x. By setting the numerical values in this way, even the beginner has a correct answer rate of about 80% or more, which is a level of correct answers that is generally regarded as almost successful.

  Note that the value of x may be stored in advance in a memory (not shown) or the like in the storage unit 3 or the determination unit 13. Or you may set arbitrarily the preferable numerical value according to the skill level of the operator of a double discrimination, the kind of stock, etc. by user input from the operation input part 2. FIG. The same applies to the value of y, the value of z, the value of α, and the value of x ′ described later.

  When the determination unit 12 determines that the specific target seedling is a double blooming seedling, the discrimination device 10 distinguishes the specific target seedling as a double blooming seedling and causes the display unit 1 to display the discrimination result. On the other hand, when the determination unit 12 determines that the specific target seedling is a single blooming seedling, the discrimination device 10 discriminates that the specific target seedling is a single blooming seedling and causes the display unit 1 to display the discrimination result. .

<Difference method of double-flowering seedlings>
As shown in FIG. 2, first, in step 101 (hereinafter abbreviated as “S101”), a TIF image is generated by the imaging apparatus 500, and the image analysis unit 11 receives the TIF image from the imaging apparatus 500, whereby S102. Proceed to In S102 (image analysis step (information acquisition step)), the image analysis unit 11 converts the TIF image into the CIELAB color space and divides the image into L ^* , a ^*, and b ^* images, and the process proceeds to S103.

In S103 (image analysis step), the image analysis unit 11 performs binarization processing on the TIF image converted into the CIELAB color space, and then sets the L ^* value, a ^* value, and b for each of a plurality of target seedlings. ^* Calculate the value and proceed to S104. In S104 (image analysis step), the image analysis unit 11 calculates and obtains a cotyledon color difference ΔE based on the L ^* value, a ^* value, and b ^* value of the cotyledon for each of the plurality of target seedlings, and S105. Proceed to The image analysis unit 11 transmits the cotyledon color difference ΔE relating to all target seedlings to the determination unit 12.

  In S105 (determination step), the determination unit 12 determines, for each of a plurality of target seedlings, whether the cotyledon color difference ΔE is included in the upper x% of the cotyledon color differences ΔE of all target seedlings. To do. When it determines with YES (henceforth "Y") in S105, the determination part 12 determines that the specific object seedling used as the determination object is a double bloom seedling, and progresses to S106.

  In S106, the discrimination device 10 finally discriminates that the specific target seedling determined to be the double blooming seedling by the determination unit 12 is the double blooming seedling. The discrimination device 10 transmits the discrimination result to the display unit 1 for display.

  On the other hand, when it determines with NO (it abbreviates as "N" hereafter) by S105, the determination part 12 determines that the specific object seedling used as the determination object is a single bloom seedling, and progresses to S107. In S107, the discrimination device 10 discriminates the specific target seedling that has been determined to be a single blooming seedling by the determination unit 12 as a single blooming seedling. The discrimination device 10 transmits the discrimination result to the display unit 1 for display.

  When all of the target seedlings have been processed in S106 or S107, the eight-fold discrimination by the discrimination device 10 is completed.

<Modification>
In the octet discrimination by the discrimination device 10, parameters other than the cotyledon color difference ΔE may be used. For example, the image analysis unit 11 calculates and acquires the cotyledon saturation C ^* (color characteristic value) instead of the cotyledon color difference ΔE, and the determination unit 12 determines whether the double-flowered seedling is based on the cotyledon saturation C ^* . You may decide whether or not. Saturation is a measure of color vividness.

Specifically, the double discrimination may be performed by a method as shown in FIG. Note that the processes of S201 to S203 are the same as the processes of S101 to S103 in the flowchart of FIG. Also, the processes of S206 and S207 are the same as the processes of S106 and S107 in the flowchart of FIG. However, in S203 (image analysis step), the image analysis unit 11 does not calculate the L ^* value.

In S204 (image analysis step), the image analysis unit 11 calculates and obtains the cotyledon saturation C ^* based on the a ^* value and the b ^* value of the cotyledon for each of the plurality of target seedlings, and proceeds to S205. . The cotyledon saturation C ^* is calculated using Equation 2 below. In addition, the image analysis unit 11 transmits the cotyledon saturation C ^* related to all target seedlings to the determination unit 12.
(Formula 2)

In S205 (determination step), for each of the plurality of target seedlings, the determination unit 12 has a cotyledon saturation C ^* higher than y% of the cotyledon saturation C ^* of all target seedlings (first range). It is determined whether it is included. Here, a numerical value of at least 50 or more is used as the value of y. By setting the numerical values in this way, even the beginner has a correct answer rate of about 80% or more, which is a level of correct answers that is generally regarded as almost successful.

  Also, for example, the image analysis unit 11 calculates and acquires the cotyledon leaf area s (size value) instead of the cotyledon color difference ΔE, and the determination unit 12 determines whether or not the seedling is a double-flowered seedling based on the cotyledon leaf area s. May be determined. Leaf area is the leaf surface area. In each of the embodiments below, the cotyledon leaf area s is defined as the area of the region surrounded by the outer periphery of the cotyledon when the cotyledon of the target seedling is viewed in plan.

  Specifically, the double discrimination may be performed by a method as shown in FIG. Note that the processes of S301, S304, and S305 are the same as the processes of S101, S106, and S107 in the flowchart of FIG. Further, the processing corresponding to S102 and S103 in the flowchart of FIG. 2 is not performed by the method shown in FIG.

  In S302 (image analysis step), the image analysis unit 11 analyzes a binarized image (size value information) generated by performing binarization processing on the TIF image received from the imaging apparatus 500, thereby obtaining a plurality of objects. The cotyledon leaf area s is acquired for each of the seedlings, and the process proceeds to S303. The image analysis unit 11 transmits the cotyledon leaf area s related to all target seedlings to the determination unit 12.

  In S303 (determination step), the determination unit 12 includes, for each of the plurality of target seedlings, the cotyledon leaf area s included in the upper z% (first range) of the cotyledon leaf area s related to all target seedlings. It is determined whether or not. Here, a numerical value of at least 40 or more is used as the value of z. By setting the numerical values in this way, even the beginner has a correct answer rate of about 80% or more, which is a level of correct answers that is generally regarded as almost successful.

  Also, for example, although not shown, the image analysis unit 11 calculates and acquires the cotyledon hue angle h (color characteristic value) instead of the cotyledon color difference ΔE, and the determining unit 12 determines the cotyledon hue angle h based on the cotyledon hue angle h. You may decide whether it is a double blooming seedling. The hue angle represents the hue as an angle.

Specifically, the image analysis unit 11 may calculate and obtain the cotyledon hue angle h based on the L ^* value, the a ^* value, and the b ^* value for each of a plurality of target seedlings. The cotyledon hue angle h is calculated using Equation 3 below.
(Formula 3)

  For each of a plurality of target seedlings, the determination unit 12 includes the cotyledon hue angle h received from the image analysis unit 11 in the lower α% (first range) of the cotyledon hue angles h of all target seedlings. It is determined whether or not. Here, a value of at least 60 or more is used as the value of α. By setting the numerical values in this way, even the beginner has a correct answer rate of about 80% or more, which is a level of correct answers that is generally regarded as almost successful.

  Note that the image analysis method and the method for determining double-flowering seedlings described in the present embodiment and the above-described modifications are merely examples, and the present invention is not limited to these cases. For example, the eight parameters may be identified using any two of the four parameters (color difference, saturation, leaf area, hue angle) described above. Alternatively, the eight-fold discrimination may be performed using parameters other than the four parameters described above that represent either the cotyledon color characteristic value or the cotyledon size value. Furthermore, the above-described four parameters, information on the color characteristic values that are the basis of these parameters, and information on the size values do not necessarily have to be acquired by image analysis.

  In other words, the discrimination device according to one aspect of the present invention includes a configuration for acquiring at least one of some information representing the color characteristic value of the cotyledon and some information representing the size value of the cotyledon from the images of a plurality of target seedlings. It only has to be. In addition, the discrimination device according to one aspect of the present invention includes whether or not any color characteristic value related to the cotyledon is included in the predetermined numerical range, and whether any size value related to the cotyledon is included in the predetermined numerical range. It suffices to have a configuration for determining at least one of whether or not.

Here, there is a possibility that a beginner can easily perform double discrimination even if a parameter representing a cotyledon shape is used in addition to a parameter representing a cotyledon color characteristic value and a parameter representing a cotyledon size value. For example, the case where the circularity of the cotyledon is used as a parameter representing the shape of the cotyledon can be considered. “Circularity” is an index indicating the roundness of the shape, and it is considered that the circularity is large (round) in single-flowering seedlings and the circularity is small (elongated) in double-flowering seedlings. The circularity is calculated using Equation 4 below.
(Formula 4)
Circularity = 4 × leaf area of cotyledon / (π × maximum diameter of cotyledon ² )
By calculating the circularity of the stock seedlings to be differentiated by image analysis or the like, and sequentially selecting the seedlings from the seedlings with lower circularity as double-flowering seedlings, even a beginner may be able to easily perform the 8-fold differentiation.

[Embodiment 2]
The information processing apparatus 200 illustrated in FIG. 1 is different from the first embodiment in that the image analysis unit 11 of the discrimination apparatus 10 calculates and acquires the cotyledon saturation C ^* and the cotyledon leaf area s in addition to the cotyledon color difference ΔE. Different from the information processing apparatus 100 according to FIG. In addition, the information processing apparatus 200 determines whether the determination unit 12 of the discrimination apparatus 10 includes the cotyledon color difference ΔE, the cotyledon saturation C ^*, and the cotyledon leaf area s within a predetermined numerical range. This is also different from the information processing apparatus 100.

<Difference method of double-flowering seedlings>
In the flowchart shown in FIG. 5, the processes of S401, S403, and S404 are the same as the processes of S101 to S103 in the flowchart of FIG. Further, the process of S402 is the same as the process of S302 in the flowchart of FIG. Furthermore, the processes of S407 and S408 are the same as the processes of S106 and S107 in the flowchart of FIG.

Note that after calculating and acquiring the cotyledon leaf area s in S402 (image analysis step), the image analysis unit 11 temporarily returns the binarized image to the original TIF image. Then, the restored TIF image is converted into the CIELAB color space and divided into L ^* , a ^*, and b ^* images (S403 (image analysis step)).

In S405 (image analysis step), the image analysis unit 11 calculates a cotyledon color difference ΔE / saturation C ^* based on the L ^* value, a ^* value, and b ^* value of the cotyledon for each of a plurality of target seedlings. Acquire and proceed to S406. The image analysis unit 11 transmits the color difference ΔE, saturation C ^*, and leaf area s of the cotyledons related to all target seedlings to the determination unit 12.

In S406 (determination step), the determination unit 12 determines whether (i) the color difference ΔE is included in the upper x ′% of all the color differences ΔE of the cotyledons related to the specific target seedling, (ii) Whether the degree C ^* is included in the top x ′% of all the saturations C ^* , and (iii) whether the leaf area s is included in the top 10% of all the leaf areas s Determine whether or not.

  Here, a numerical value of at least 40 or more is used as the value of x ′. By setting the numerical values in this way, even the beginner has a correct answer rate of about 80% or more, which is a level of correct answers that is generally regarded as almost successful.

When any one or more of the cotyledon color difference ΔE, saturation C ^*, and leaf area s is included in the numerical range of (i) to (iii), the determination unit 12 determines the specific target that is the determination target. It is determined that the target seedling is a double blooming seedling. On the other hand, when none of the cotyledon color difference ΔE, saturation C ^*, and leaf area s is included in the numerical range of (i) to (iii), the determining unit 12 causes the specific target seedling to bloom in a single bloom. Decide that it is a seedling.

The above-described determination process by the determination unit 12 is merely an example, and is not limited to this case. For example, even if any two or more of cotyledon color difference ΔE, saturation C ^*, and leaf area s are included in the numerical range of (i) to (iii) above, even if it is determined that the seedling is an double blooming seedling Good. Alternatively, when all of the cotyledon color difference ΔE, saturation C ^*, and leaf area s are included in the numerical ranges of (i) to (iii), it may be determined that the seedling is a double-flowering seedling.

[Embodiment 3]
In the information processing apparatus 300 illustrated in FIG. 1, the image analysis unit 11 of the discrimination apparatus 10 calculates and acquires the cotyledon saturation C ^* , the cotyledon leaf area s, and the cotyledon hue angle h in addition to the cotyledon color difference ΔE. This is different from the information processing apparatuses 100 and 200 according to the first and second embodiments. In the information processing apparatus 300, the determination unit 12 of the discrimination apparatus 10 includes the cotyledon color difference ΔE, the cotyledon saturation C ^* , the cotyledon leaf area s, and the cotyledon hue angle h in predetermined numerical ranges. It is also different from the information processing apparatuses 100 and 200 in that it is determined whether or not.

<Difference method of double-flowering seedlings>
In the flowchart shown in FIG. 6, the processes in S501, S503, and S504 are the same as the processes in S101 to S103 in the flowchart in FIG. Further, the processing of S502 is the same as the processing of S302 in the flowchart of FIG. Furthermore, the processing of S508 and S509 is the same as the processing of S106 and S107 in the flowchart of FIG.

In S505 (image analysis step), for each of the plurality of target seedlings, the image analysis unit 11 uses the cotyledon color difference ΔE, saturation C ^*, and hue angle based on the L ^* value, a ^* value, and b ^* value of the cotyledon. h is calculated and acquired, and the process proceeds to S506. The image analysis unit 11 transmits the cotyledon color difference ΔE, the saturation C ^* , the leaf area s, and the hue angle h of all the target seedlings to the determination unit 12.

In S506 (determination step), the determination unit 12 determines whether (I) the color difference ΔE is included in the top 30% of all the color differences ΔE of the cotyledons related to the specific target seedling, or (II) saturation Whether C ^* is included in the top 30% of all chroma C ^* , (III) whether leaf area s is included in the top 20% of all leaf areas s, and (IV) It is determined whether the hue angle h is included in the lower 30% of all the hue angles h.

  By setting the numerical value range as described above, even for beginners, the correct answer rate for the double discrimination becomes about 80% or more, and the correct answer rate is generally regarded as almost successful.

When it is determined as Y in S506, in other words, any one or more of the cotyledon color difference ΔE, the saturation C ^* , the leaf area s, and the hue angle h are included in the numerical range of the above (I) to (IV). If YES, go to S507.

  In S507 (determination step), the determination unit 12 determines whether (V) the leaf area s of the cotyledon related to the specific target seedling is included in 80% or less of the average value M of all the leaf areas s. To do. Furthermore, (VI) it is determined whether or not the color difference ΔE of the cotyledon related to the specific target seedling is included in the lower 40% or less of the color difference ΔE of all the cotyledons, and the color of the cotyledon related to the specific target seedling It is determined whether or not the angle h is included in the upper 40% or more of the hue angles h of all the cotyledons.

  In addition, even if it does not perform the process of S507, the correct answer rate of the double discrimination by a beginner can be maintained at about 80% or more. However, the correct answer rate can be further increased by further performing the processing of S507.

If N is determined in S507, in other words, if none of the cotyledon color difference ΔE, saturation C ^*, and leaf area s is included in the numerical ranges of (V) and (VI), the process proceeds to S508.

On the other hand, when N is determined in S506, in other words, none of the cotyledon color difference ΔE, saturation C ^* , leaf area s, and hue angle h is included in the numerical range of the above (I) to (IV). If YES, the process proceeds to S509. If it is determined as Y in S507, in other words, at least the cotyledon leaf area s is included in the numerical range of (V), or the cotyledon color difference ΔE and the hue angle h are numerical values of (VI). If it is included in the range, the process proceeds to S509.

The above-described determination process by the determination unit 12 is merely an example, and is not limited to this case. For example, when any two or more of cotyledon color difference ΔE, saturation C ^* , leaf area s, and hue angle h are included in the numerical range of (I) to (IV), the process of S507 is performed. May be. Alternatively, when all of the cotyledon color difference ΔE, saturation C ^*, and leaf area s are included in the numerical ranges of (V) and (VI), it may be determined that the seedling is a double-flowering seedling.

〔Example〕
After seed seeds were seeded one by one in a cell tray or pot, the seedlings in a state where the cotyledons developed and the main leaves began to appear were subjected to double discrimination according to one embodiment of the present invention. Hereinafter, various processes performed at the time of the double discrimination and the discrimination results will be described.

  Three kinds of seeds of iron marine, iron cherry and iron white were used as stock seeds. In the following description, iron marine seedlings are “target seedlings of Examples 1 and 4”, iron cherry seedlings are “target seedlings of Example 2”, and iron white is “target seedlings of Example 3”. And The outline of each embodiment is as follows.

  The target seedlings of Example 1 were sown 200 seeds per cell tray and sown on a total of 6 cell trays (total number of germinated seedlings: 1161 strains). This sowing operation was performed over two days on February 3, 2017 and February 10, 2017. The image analysis work was performed using ImageJ (manufactured by NIH, free image analysis software) installed in a notebook personal computer, which will be described later, and the keyboard work and mouse operation of the notebook personal computer.

The target seedlings of Example 2 were sown 200 seeds per cell tray and sown on a total of three cell trays (total number of germinated seedlings: 576 strains). This sowing work was carried out on August 21, 2017, August 28 of the same year, and September 3 of the same year. In the image analysis work, the TIF image binarization process and the calculation of the L ^* value, a ^* value, and b ^* value are automatically performed by the ImageJ macro, and the subsequent processing is performed by the keyboard operation and mouse operation of the notebook computer. went.

  The target seedlings of Example 3 were sown 200 seeds per cell tray and sown on a total of three cell trays (total number of germinated seedlings: 584 strains). The sowing work and the image analysis work are the same as those of the target seedling of Example 2.

  The target seedlings of Example 4 were sown one by one in a 9 cm pot placed on a tray. The pots were arranged in a matrix of 6 × 4 (24 in total) on the tray. A total of 19 trays were seeded with seedlings, and the total number of germinated seedlings was 423. This sowing work was carried out on August 29, 2017, September 13 of the same year, and November 3 of the same year. The image analysis work is the same as that of the target seedling of Example 1, and the method of determining the double blooming seedling is the same as the method (vi) described later.

<Generation of TIF images>
For the target seedlings of Examples 1 to 3, one seeded cell tray was placed in a dark room and imaged from above with the same field of view. For the target seedlings of Example 4, one seeded tray was placed in the same darkroom and imaged from above with the same field of view. For all of Examples 1 to 4, a recording lamp (National PRF-500WB, color temperature 5500K) was used for illumination.

  In all of Examples 1 to 4, a digital camera (Nikon D3200) was used as the imaging device 500, and imaging was performed with aperture priority auto. The settings of the digital camera were: aperture value: F11, ISO sensitivity: 100, focus mode: AF-S single AF servo, white balance: clear sky, metering: multi-pattern metering, flash: none, image quality mode: RAW.

The RAW format initial image captured by the digital camera was developed with RAW development software Capture NX-D ( Nikon Corporation ) to generate a TIF image (8-bit, TIF format initial image). The development settings are as follows: color system: sRGB color, white balance of Example 1: high color rendering fluorescent lamp (color temperature 5500K), white balance of Examples 2, 3 and 4: clear sky (color temperature 5500K), picture Control: Neutral.

<Image analysis>
For convenience of explanation, only the target seedlings of Example 1 are shown in FIGS. The above TIF image was analyzed in the following procedure using ImageJ installed in a notebook personal computer (D540S: manufactured by ASUS, information processing apparatus 100/200). The TIF image before image analysis and the basic screen of ImageJ are displayed in the form and arrangement as shown in FIG. 7 on the display screen of the notebook computer. The “seed image” in FIG. 7 corresponds to a TIF image before image analysis.
(1) As shown in FIG. 8, “Dark background” was selected in “Threshold Color” displayed on the left side of the display screen of the notebook computer. Thereafter, in order to separate and handle the image portion of the target seedling and the background image portion, the image portion of the target seedling was segmented by the Hue value, the Saturation value, and the Brightness value, and the TIF image was converted into a binary image. This binarized image is displayed on the display screen of the notebook personal computer in a manner and arrangement as shown in FIG.

Here, the hue value corresponds to a hue threshold value, the saturation value corresponds to a saturation threshold value, and the brightness value corresponds to a lightness threshold value.
(2) The above binarized image was activated, and “Original” in “Threshold Color” was selected (see FIG. 8). Thereafter, the original TIF image (RGB image) was converted to CIELAB color space by RGB to CIELAB (plug-in) and divided into L ^* , a ^*, and b ^* images.

As shown in FIGS. 9A to 9C, the L ^* , a ^*, and b ^* images are displayed on the right side of the display screen of the notebook computer. In all these three cases, the original TIF image is displayed on the left side of the notebook screen.
(3) In “Analyze Particles”, “Exclude on edges, Include holes” was selected and set to “size: 1500-Infinity”. Thereafter, image analysis was performed, and for each of the target seedlings according to Examples 1 to 4, the threshold value between the target seedling and the background in the image and the leaf area s of the cotyledon were calculated. In all of Examples 1 to 4, the threshold value was always calculated, and the leaf area s of the cotyledon was appropriately calculated as necessary.
(4) The above binarized image was activated again, “Redirect to” in “Set Measurements” was selected, and each image of L ^* , a ^* and b ^* was designated in order. Thereafter, “Analyze Particles” was selected and image analysis was performed, and the L ^* value, a ^* value, and b ^* value of the cotyledons were calculated for each of the target seedlings according to Examples 1 to 4.

The calculated L ^* value, a ^* value, and b ^* value of the cotyledon are displayed on the left side of the display screen of the notebook computer as shown in FIGS. Further, the designated L ^* image is displayed on the right side of the L ^* value calculation result as shown in FIG. The same applies to the designated images of a ^* and b ^* (see (b) and (c) of FIG. 10).
(5) For each of the 200 target seedlings according to Examples 1 to 3, the cotyledon color difference ΔE, the chroma C ^*, and the leaf area s were calculated by the above-described formulas 1 and 2. For each of the target seedlings according to Example 1, the hue angle h was further calculated by Equation 3 described above.

For each subject seedlings according to Examples 1 to 4, to represent the respective calculation results of the color difference Delta] E · chroma C ^* · leaf area s · hue angle h of the cotyledon with histones gram, for example, as shown in FIG. 11 It became a histogram. The histogram in FIG. 11 is generated based on the TIF image in FIG. 7, and the solid line graph in the figure is the double-flowered seedling and the broken line graph is the single-flowered seedling.

  In addition, the display mode on the screen (display unit) of the notebook computer shown in FIGS. 7 to 10 and FIGS. 13 and 19 to be described later is merely an example, and is not limited to these cases. The display mode may be changed by appropriately changing the settings of the notebook computer or ImageJ. Alternatively, the above-described display mode may be changed by using an information processing device (such as a tablet terminal) other than a notebook computer or using image analysis software other than ImageJ.

<Difference method and results>
For each of the target seedlings according to Examples 1 to 3, (i) octet discrimination using only the cotyledon color difference ΔE, (ii) octet discrimination using only the cotyledon saturation C ^* , and (iii) cotyledon leaf area An octet discrimination using only s and (iv) an octet discrimination using cotyledon color difference ΔE, saturation C ^* and leaf area s were performed.

For the target seedling according to Example 1, (v) double discrimination using only the cotyledon hue angle h, and (vi) cotyledon color difference ΔE, saturation C ^* , leaf area s, hue angle h are used. There was a double discrimination. About the target seedling which concerns on Example 4, the double discrimination by the method of (vi) was performed.

  In addition, the worker who performed each of the above-described eight-fold discrimination is a woman (gender) having an experience of 5 months, and corresponds to a “beginner” of the eight-fold discrimination. Further, in each embodiment, the eight-fold discrimination is performed for each image (one photograph), and the discrimination results (tables and graphs) shown in FIGS. 12 to 19 are all objects projected across a plurality of images. Calculated and generated based on the total and average values of the seedlings.

(Difference method)
In the discrimination method (i), the cotyledon color difference ΔE related to a specific target seedling is the upper x% (x = 10, 20, 30, 40, 50, 60) of the cotyledon color difference ΔE related to all target seedlings. ) To determine whether it is a double-flowered seedling. In the discrimination method of (ii), the cotyledon saturation C ^* related to a specific target seedling is the top y% (y = 10, 20, 30, 40) of the cotyledon saturation C ^* related to all target seedlings. , 50, 60) to determine whether it is a double-flowering seedling. In the identification method of (iii), the cotyledon leaf area s related to the specific target seedling is the upper z% (z = 10, 20, 30, 40, 50) of the cotyledon leaf area s related to all the target seedlings. 60) and whether it is a double-flowered seedling was determined.

In the discrimination method of (iv), (i) the color difference ΔE is included in the upper x ′% of all the color differences ΔE of the cotyledons related to the specific target seedling, and (ii) the saturation C ^* is It is determined whether or not it is included in the top x ′% of all the saturations C ^* , and (iii) whether or not the leaf area s is included in the top 10% of all the leaf areas s, It was decided whether it was a double blooming seedling. Here, for the target seedlings of Example 1, x ′ = 20, 30, 40, 50, 55, and for the target seedlings of Examples 2 and 3, x ′ = 10, 20, 30, 40, 50 did.

In the discrimination method of (v), the hue angle h of the cotyledons related to a specific target seedling is the lower α% (α = 10, 20, 30, 40, 50) of the hue angles h of the cotyledons related to all the target seedlings. 60) and whether it is a double-flowering seedling was determined.

In the identification method of (vi), (I) color difference ΔE is included in the top 30% of all color differences ΔE of cotyledons related to a specific target seedling, and (II) chroma C ^* is all chroma C ^* whether or not included in the top 30% of the, whether or not included in the top 20% of (III) leaf area s all leaf area s, and (IV) hue It was determined whether or not the angle h was included in the lower 30% of all the hue angles h.

Next, when any one or more of the cotyledon color difference ΔE, the saturation C ^* , the leaf area s, and the hue angle h are included in the numerical range of (I) to (IV) above, It is determined whether or not the cotyledon leaf area s related to the target seedling is included in 80% or less of the average value M. Furthermore, (VI) it is determined whether or not the color difference ΔE of the cotyledon related to the specific target seedling is included in the lower 40% or less of the color difference ΔE of all the cotyledons, and the color of the cotyledon related to the specific target seedling It was determined whether or not the angle h was included in the upper 40% or more of the hue angles h of all cotyledons, and it was determined whether or not the seedling was a double blooming seedling.

(Result of discrimination)
First, with respect to the target seedling of Example 1, (a) in FIG. 12 shows the discrimination results when the double discrimination is performed by the methods (i) to (iii) and (v). As shown in FIG. 12 (a), in the discrimination methods (i), (ii) and (v), the correct answer rate was about 80% or more for all values of x, y and α = 60 or more. . In the identification method of (iii), the correct answer rate was about 80% or more when z = 30 or more.

  FIG. 12B shows the discrimination result when the double discrimination is performed by the method (iv). As shown in FIG. 12 (b), the correct answer rate was about 80% or more for all values of x ′ = 55 or more. FIG. 13 shows the discrimination results when the double discrimination is performed by the method (vi). FIG. 13 shows the discrimination result of one predetermined tray among the six trays. As shown in FIG. 13 (a), when the number of target seedlings (seeds surrounded by circles in the figure) determined to be double-flowering seedlings by the method of (vi) was 75, As shown in b), there were 69 target seedlings that were actually double-flowered seedlings (seeds marked with an asterisk in the figure). That is, the correct answer rate was 93%. The other five trays were similarly marked with a circle and a star. As a whole, the seedling selection rate was 39%, and the correct answer rate was 94%.

  FIG. 14 shows the relationship between the seedling selection rate and the correct answer rate when the double discrimination is performed by each of the methods (i) to (iv). The seedling selection rate is the total number of target seedlings (Example 1: 1161 strains, Example 2: 576 strains, Example 3). : 584 shares), and then multiplied by 100. In general, the numerical value of the minimum seedling selection rate required in consideration of productivity is about 30%. As shown in FIG. 14, the correct answer rates at the seedling selection rate of 30% are (i) to (iv ) Over 80% in all methods.

  Next, about the object seedling of Example 2, the discrimination result when performing double discrimination by each method of (i)-(iii) is shown to (a) of FIG. As shown in FIG. 15A, in the discrimination method (i), the correct answer rate was about 80% or more when x = 40 or more. In the discrimination method (ii), the correct answer rate was about 80% or more when y = 40 or more. In the discrimination method (iii), the correct answer rate was about 80% or more when z = 30 or more.

  FIG. 15B shows the result of discrimination when the double discrimination is performed by the method (iv). As shown in FIG. 15B, when x ′ = 40 or more, the correct answer rate exceeded 80%.

  FIG. 16 shows the relationship between the seedling selection rate and the correct answer rate when the double discrimination is performed by each method (i) to (iv). As shown in FIG. 16, the correct answer rate at the seedling selection rate of 30% exceeded 80% in all the methods (i) to (iv) as in Example 1.

  Next, with respect to the target seedling of Example 3, (a) in FIG. 17 shows the discrimination results when the double discrimination is performed by the methods (i) to (iii). As shown in FIG. 17A, in the discrimination method (i), the correct answer rate is about 80% or more when x = 50 or more. In the discrimination method (ii), the correct answer rate was about 80% or more when y = 30 or more. In the discrimination method (iii), the correct answer rate was about 80% or more when z = 40 or more.

  FIG. 17B shows the discrimination result when the double discrimination is performed by the method (iv). As shown in FIG. 17B, when x ′ = 40 or more, the correct answer rate was about 80% or more.

  FIG. 18 shows the relationship between the seedling selection rate and the correct answer rate when the double discrimination is performed by each method of (i) to (iv). As shown in FIG. 18, the correct answer rate at the seedling selectivity of 30% exceeded 85% in the methods (i) and (iv), and was almost 85% in the method (iii). On the other hand, in the method (ii), the correct answer rate at a seedling selection rate of 30% was less than 80%.

  Next, with respect to the target seedling of Example 4, FIG. 19 shows the discrimination result when the double discrimination is performed by each method of (vi). FIG. 19 shows the discrimination result of one predetermined tray out of 19 trays. The seedlings surrounded by circles in (a) of FIG. 19 were target seedlings determined to be double-flowering seedlings by the method of (vi). On the other hand, the seedlings marked with an asterisk in FIG. 19B were target seedlings that were actually double blooming seedlings. The other 18 trays were similarly marked with a circle and a star. As a whole, the seedling selection rate was 34%, and the correct answer rate was 96%.

(Discussion)
When comparing the discrimination results of each method of (i) to (iv) for each of Examples 1 to 3, the selectivity is increased when the target seedling of Example 1 is subjected to double discrimination by the method of (iii). The degree of decrease in the correct answer rate at that time became the most rapid (see FIG. 14). In addition, when the target seedling of Example 3 was subjected to double discrimination by the method (ii), the correct answer rate changed at the lowest level (see FIG. 18).

  On the other hand, when the target seedling of Example 3 was subjected to double discrimination by the method (ii), the degree of decrease in the correct answer rate when the seedling selection rate was increased was most gradual (see FIG. 18). Moreover, although the correct answer rate at 30% seedling selection rate is less than 80%, there is a high possibility of error from the tendency of the correct answer rate to decline, and the actual result is about 78% at 30% seedling selection rate. The correct answer rate is obtained (see FIG. 18).

  Regarding the method (v), since the correct answer rate at α = 60 in the target seedling of Example 1 is 78% (see FIG. 12A), at least the seedling selection rate in the target seedling of Example 1 It is presumed that the desired correct answer rate can be obtained even if the value is increased. However, the target seedlings of Examples 2, 3 and 4 are not subjected to the double discrimination by the method (v), and there is no data on the correct answer rate. Therefore, it cannot be said that the desired correct answer rate can be obtained even if the seedling selection rate is increased for all target seedlings.

  Regarding the method (vi) using a plurality of parameters, in the target seedlings of Examples 1 and 4, the correct answer rate exceeded 90% in the two cases as described above. In these cases, since the seedling selection rates were 34% and 39%, it was inferred that the desired correct answer rate could be sufficiently obtained even if the seedling selection rate was increased at least for the target seedlings of Examples 1 and 4. Is done. However, since there is no data on the correct answer rate of the target seedlings of Examples 2 and 3, it cannot be said that the desired correct answer rate can be obtained even if the seedling selection rate is increased for all target seedlings.

  From these facts, among the methods (i) to (iii) and (v) using a single parameter that is easier for beginners to make an eight-fold discrimination, the methods (i) and (ii) are the most reliable.・ It is thought that a high correct answer rate can be obtained. Therefore, among the methods (i) to (iv), it is considered to be most practical to perform the eight-fold discrimination by the method (i) or (ii).

  Next, with respect to the method (iv) using a plurality of parameters, in all of the target seedlings of Examples 1 to 3, a higher correct answer rate was obtained even when compared with the methods (i) to (iii). (See FIGS. 14, 16 and 18).

  Specifically, in the target seedling of Example 1, the correct answer rate is about 0.5% in the range of the seedling selection rate of 30% to 60% than in the case of the double discrimination by the methods (i) and (ii). Increased by about 2.4% (see FIG. 14). In addition, in the target seedling of Example 2, the correct answer rate is increased by about 2.0% to about 2.5% in the range of the seedling selection rate of 30% to 50%, compared with the case of the double discrimination by the method of (i), The correct answer rate increased by about 3.5% compared to the case of the double discrimination by the method (ii) (see FIG. 16).

  From these facts, although it takes a little more effort than the methods (i) to (iii) and (v) using a single parameter, if the emphasis is on improving the productivity and the correct answer rate, (iv ) Method is considered to be most effective.

[Comparative Example]
With respect to the same target seedlings sown on August 21, 2017, the first novice visually discriminates the octet, the expert witnesses the octet discrimination, and the discriminating method according to one embodiment of the present invention (No. Conducted by 1 beginner). Moreover, the double discrimination (implemented by a second beginner different from the first beginner) by the above-described discrimination method was performed on the target seedlings sown in February 2017. The discrimination results are shown in FIG. Here, the “skilled person” of the eight-fold discrimination can be exemplified by an eight-fold discrimination worker having three or more years of experience.

  Note that “beginner” in the figure corresponds to the first beginner, and “image processing” corresponds to the double discrimination in the discrimination method according to one aspect of the present invention. Further, the first beginner has received guidance from an expert in advance, and the second beginner has not received guidance from the expert.

  As shown in FIG. 20, for the target seedlings sown in February 2017, the second novice is in charge of the present invention despite the fact that the number of seedlings discriminated eight times from the target seedlings sown on August 21, 2017 is considerably large. Even if the discrimination method according to one aspect was used for the eight-fold discrimination, the same 95% correct answer rate as that of the expert could be obtained. In addition, regarding the target seedlings sown on August 21, 2017, the correct answer rate of 87%, which is regarded as successful when the first beginner performs an eight-fold discrimination with the discrimination method according to one aspect of the present invention. Could get.

  Therefore, regardless of the presence or absence of guidance from a skilled person or the presence or absence of prior knowledge regarding double discrimination, any beginner can be regarded as almost successful according to the discrimination method according to one aspect of the present invention. It is thought that the correct answer rate of the level to be deceived can be obtained.

  In addition, about the target seedlings sown on August 21, 2017, the correct answer rate is 84% according to visual double discrimination by the first beginner. Such a result is presumably due to the fact that the first beginner was previously instructed by a skilled person. If there is no instruction from a skilled person, the correct answer rate will naturally decrease.

[Example of software implementation]
The control block (particularly the image analysis unit 11 and the determination unit 12) of the discrimination device 10 may be realized by a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like, or may be realized by software. Good.

  In the latter case, the discrimination device 10 includes a computer that executes instructions of a program that is software that realizes each function. The computer includes, for example, one or more processors and a computer-readable recording medium storing the program. In the computer, the processor reads the program from the recording medium and executes the program, thereby achieving the object of the present invention. As the processor, for example, a CPU (Central Processing Unit) can be used. As the recording medium, a “non-temporary tangible medium” such as a ROM (Read Only Memory), a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used. Further, a RAM (Random Access Memory) for expanding the program may be further provided. The program may be supplied to the computer via an arbitrary transmission medium (such as a communication network or a broadcast wave) that can transmit the program. Note that one embodiment of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission.

[Additional Notes]
The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention.

10 Identification Device 11 Image Analysis Unit (Information Acquisition Unit)
12 Deciding part 100, 200, 300 Information processing device ΔE Cotyledon color difference C ^* Cotyledon saturation s Cotyledon leaf area h Cotyledon hue angle

Claims (11)

A discriminating apparatus for discriminating the double flower seedlings from a previous plurality of stock seedlings to differentiate double flower seedlings,
Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition unit;
The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the order of the color characteristic values for each of the plurality of stock seedlings is within a predetermined first range. It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining at least one of whether or not it is included and whether or not the order of the size values is included in a predetermined second range a determination unit which comprises a,
The information acquisition unit acquires a color difference of the cotyledon from the color characteristic value information,
The determination unit determines whether the color difference of a specific seedling of the stock is included in at least the upper 50% of all the color differences in the first range. apparatus. The information acquisition unit acquires the cotyledon saturation from the color characteristic value information,
The determination unit determines whether or not the saturation of a specific stock seedling is included in at least the top 50% of all the saturations in the first range. The identification device according to claim 1 . The information acquisition unit acquires the hue angle of the cotyledon from the color characteristic value information,
The determination unit determines whether or not the hue angle of a specific seedling of the stock is included in at least the lower 60% of all the hue angles in the first range. The identification device according to claim 1 or 2 . The information acquisition unit acquires a leaf area of the cotyledon from the size value information,
The determining unit determines whether or not the leaf area of a specific seedling of the stock is included in at least the top 40% of all the leaf areas that are the second range. The identification device according to any one of claims 1 to 3 . A discrimination device for distinguishing the double-flowering seedlings from a plurality of stock seedlings before differentiating the double-flowering seedlings,
Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition unit;
The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the order of the color characteristic values for each of the plurality of stock seedlings falls within a predetermined first range. It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining at least one of whether or not it is included and whether or not the size value order is included in a predetermined second range And a determination unit to
The information acquisition unit acquires the cotyledon color difference and the cotyledon saturation from the color characteristic value information, acquires the cotyledon leaf area from the size value information,
The determination unit determines whether (i) the color difference of a specific stock seedling is included in at least the top 40% of all the color differences in the first range, (ii) a specific Whether the saturation of the stock seedling is included in at least the top 40% of all the saturations in the first range, and (iii) the leaf of the particular stock seedling It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining whether or not the area is included in the top 10% of all the leaf areas in the second range Kan by device you wherein a.   A discrimination device for distinguishing the double-flowering seedlings from a plurality of stock seedlings before differentiating the double-flowering seedlings,
  Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition unit;
  The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the order of the color characteristic values for each of the plurality of stock seedlings is within a predetermined first range. It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining at least one of whether or not it is included and whether or not the order of the size values is included in a predetermined second range And a determination unit to
  The information acquisition unit acquires the cotyledon saturation from the color characteristic value information,
  The determination unit determines whether or not the saturation of a specific stock seedling is included in at least the top 50% of all the saturations in the first range. Identification device to do.   A discrimination device for distinguishing the double-flowering seedlings from a plurality of stock seedlings before differentiating the double-flowering seedlings,
  Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition unit;
  The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the order of the color characteristic values for each of the plurality of stock seedlings is within a predetermined first range. It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining at least one of whether or not it is included and whether or not the order of the size values is included in a predetermined second range And a determination unit to
  The information acquisition unit acquires the hue angle of the cotyledon from the color characteristic value information,
  The determination unit determines whether or not the hue angle of a specific seedling of the stock is included in at least the lower 60% of all the hue angles in the first range. Identification device to do.   A discrimination device for distinguishing the double-flowering seedlings from a plurality of stock seedlings before differentiating the double-flowering seedlings,
  Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition unit;
  The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the order of the color characteristic values for each of the plurality of stock seedlings is within a predetermined first range. It is determined whether or not a particular seedling of the stock is the double-flowering seedling by determining at least one of whether or not it is included and whether or not the order of the size values is included in a predetermined second range And a determination unit to
  The information acquisition unit acquires a leaf area of the cotyledon from the size value information,
  The determining unit determines whether or not the leaf area of a specific seedling of the stock is included in at least the top 40% of all the leaf areas that are the second range. Identification device to do. The information processing apparatus characterized by comprising a discrimination device according to any one of claims 1 to 8.   A program for causing a computer to function as the identification device according to claim 1, wherein the program causes the computer to function as the information acquisition unit and the determination unit. A plurality of stock seedlings before differentiating double flower seedlings a discrimination method for discriminating the double flower seedlings,
Information for acquiring at least one of color characteristic value information representing the cotyledon color characteristic value and size value information representing the cotyledon size value for each of the plurality of stock seedlings from the plurality of stock seedling images. An acquisition step;
The color characteristic values and the size values of the plurality of stock seedlings are ranked in descending order, and the color characteristic values acquired in the information acquisition step for each of the plurality of stock seedlings. ranking whether contained in the predetermined first range, and ranking of the size value acquired by the information acquisition step determines at least one of whether or not within a predetermined second range Determining whether a particular seedling of the stock is the double blooming seedling, and
The information acquisition step acquires a color difference of the cotyledon from the color characteristic value information,
The determination step determines whether or not the color difference of a specific seedling of the stock is included in at least the top 50% of all the color differences in the first range. Method.