Estimation model generation apparatus and estimation model generation method

Abstract

To estimate a growth condition of plants in a field by using a vegetation index that is suitable for estimating the growth condition.SOLUTION: An estimation model generation apparatus includes: a data augmentation unit configured to generate a computation formula for obtaining a new vegetation index by modifying an existing computation formula for calculating a vegetation index; a vegetation index calculation unit configured to calculate the vegetation index by using the computation formula for obtaining the new vegetation index generated by the data augmentation unit, based on acquired image data obtained by capturing images of a field; an estimation model generation unit configured to generate an estimation model for estimating a growth condition of plants at each position in the field, based on a relationship between a growth index value, which is determined from an evaluation item indicating the growth condition of the plants in the field, and the vegetation index calculated by the vegetation index calculation unit; and an estimation model suitability determination unit configured to determine the suitability of the estimation model based on the estimation accuracy of the estimation model generated by the estimation model generation unit.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an estimation model generating device and an estimation model generating method. [Background technology]

[0002] A vegetation index is a method for understanding the growth state of plants grown in a field. The vegetation index is an index that indicates the characteristics of light reflection by plants. The vegetation index can be calculated, for example, using the intensity of reflected light having several specific wavelengths contained in image data obtained by photographing the field. Since the vegetation index indicates the characteristics of light reflection by plants, it is used as an index for understanding the growth state of plants. There are various types of vegetation indexes, such as NDVI, NDRE, and NDWI. Patent Document 1 discloses that one of these vegetation indices (for example, NDVI) is used to evaluate the quality of the growth state of plants. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-009457 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the above-described technology evaluates multiple fields using a single vegetation index (e.g., NDVI) that is predetermined for each field. Soil colors vary, with differences depending on the region, such as black or dark soil, reddish soil, or yellow soil. Plant color may also change due to the influence of soil components. Similarly, the light reflection characteristics of light components other than visible light, which are invisible to the human eye, vary depending on the soil, and it is thought that plants affected by the soil also have different light reflection characteristics. Therefore, even if the growth conditions of plants in multiple fields are the same, the vegetation index calculated for each field may differ. Furthermore, even if the vegetation index calculated from images of multiple fields is the same, the growth conditions of plants may differ from field to field. Therefore, it is possible to accurately grasp the growth conditions of plants by using an appropriate vegetation index for each field.

[0005] The present invention has been made in consideration of the above circumstances, and its purpose is to provide an estimation model generation device and an estimation model generation method that can estimate the growth state of plants in a field using a vegetation index that is suitable for estimating the growth state of plants in a field. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems, one aspect of the present invention is an estimation model generation device that includes a data expansion unit that generates a calculation formula for calculating a new vegetation index by making changes to a calculation formula for calculating a vegetation index; a vegetation index calculation unit that calculates a vegetation index using the calculation formula for the new vegetation index generated by the data expansion unit from acquired image data obtained by photographing the field; an estimation model generation unit that generates an estimation model that estimates the growth state of plants at each position in the field from the relationship between a growth index value based on an evaluation item that indicates the growth state of plants in the field and the vegetation index calculated by the vegetation index calculation unit; and an estimation model suitability determination unit that determines the suitability of the estimation model based on the estimation accuracy of the estimation model generated by the estimation model generation unit.

[0007] Another aspect of the present invention is an estimation model generation method in which a data extension unit generates a calculation formula for calculating a new vegetation index by making changes to a calculation formula for calculating a vegetation index, a vegetation index calculation unit calculates a vegetation index using the calculation formula for the new vegetation index generated by the data extension unit from acquired image data obtained by photographing the field, an estimation model generation unit generates an estimation model that estimates the growth state of plants at each position in the field from the relationship between a growth index value based on an evaluation item that indicates the growth state of plants in the field and the vegetation index calculated by the vegetation index calculation unit, and an estimation model suitability determination unit determines the suitability of the estimation model based on the estimation accuracy of the estimation model generated by the estimation model generation unit. [Effects of the Invention]

[0008] As described above, according to the present invention, the growth state of plants in a farm field can be estimated using a vegetation index suitable for estimating the growth state of plants in the field. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a schematic block diagram showing the configuration of an estimation model generating device 1 according to an embodiment. [Figure 2] FIG. 3 is a diagram showing an example of information stored in a vegetation index information storage unit 121. [Figure 3] FIG. 2 is a diagram for explaining the processing performed by the estimation model generating unit 15. [Figure 4] 10 is a diagram for explaining the processing performed by the estimation accuracy calculation unit 17. FIG. [Figure 5] FIG. 2 is a diagram illustrating an example of information stored in an estimation model information storage unit 19. [Figure 6] 1 is a flowchart illustrating the flow of processing performed by the estimation model generating device 1. DETAILED DESCRIPTION OF THE INVENTION

[0010] An estimation model generating device according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of an estimation model generating device 1 according to the first embodiment. The estimation model generation device 1 has a field survey data acquisition unit 11, a calculation unit 12, a filter processing unit 13, a field survey point data extraction unit 14, an estimation model generation unit 15, a filter processing unit 16, an estimation accuracy calculation unit 17, an estimation model appropriateness determination unit 18, and an estimation model information storage unit 19.

[0011] The field survey data acquisition unit 11 acquires field survey data including growth index values ​​based on evaluation items that indicate the growth state of plants in the field, such as the number of stems, plant height, and leaf color of plants grown in the field. The plants grown in the field are, for example, wheat. In addition, in the field survey, the person in charge of the survey actually goes to the field and measures and obtains the growth index values ​​at multiple points in the field. A specific example of a field survey of growth index values ​​is shown below. For example, by investigating the number of stems at each of multiple locations in a field, it is possible to determine the number of stems per 1 m at each of multiple different locations. 2 The number of stems per unit area (growth index value) can be obtained. For example, leaf color is expressed by a value calculated based on the transmittance of two types of light, red light (wavelength 650 nm) and infrared light (wavelength 940 nm), irradiated onto the leaf. Leaf color can be expressed as a SPAD value, which is the amount of chlorophyll (chlorophyll content) contained in the plant's leaves. SPAD values ​​are sometimes used to understand the health of plants. SPAD values ​​can be measured using a SPAD measuring device, which is sometimes called a chlorophyll meter.

[0012] The field survey data acquisition unit 11 acquires growth index values ​​based on the results of a field survey of information that can be expressed quantitatively and qualitatively, such as the number of stems, the number of upper stems, the plant height, the water content, the water content rate, and / or the leaf color (at least one of the number of stems, the number of upper stems, the plant height, the water content, the water content rate, or the leaf color) of plants grown in the field that is the subject of the survey, at a time corresponding to the time when the acquired image data was measured. The acquired image data is acquired image data measured at the time when the field survey was conducted. From the viewpoint of improving the accuracy of estimating the growth index value, it is preferable for the field survey data acquisition unit 11 to acquire field survey data for the period when the growth stage is desired to be confirmed. The acquired image data may be data measured on a day just before or just after that. This is because it is sufficient to obtain a correlation between the acquired image data and the field survey data, and even if the days are different, the tendency of variation in the growth state of plants in the field is unlikely to change so significantly that it would affect the estimation of the number of stems. Furthermore, the growth index value may be a newly calculated value obtained by applying an arbitrary function to a single or multiple growth index values ​​obtained by a field survey. The field survey data acquisition unit 11 also includes a process for applying an arbitrary function to the growth index value when applying an arbitrary function to the growth index value obtained by the field survey to calculate a new calculated value. Specifically, an example of a process for calculating a new value from the growth index value obtained by the field survey is a process for calculating the amount of a sprayed product as a newly calculated amount from the growth index values ​​when a sprayed product is to be applied to a field from two growth index values, namely the number of stems and leaf color of a plant obtained by the field survey. An example of such a sprayed product is fertilizer sprayed on a field. When fertilizer is used as the sprayed product, the amount of sprayed product is the amount of fertilizer applied.

[0013] The field survey data is data in which, for example, an identifier (ID, hereinafter referred to as ID), the number of stems, leaf color, latitude, and longitude are associated with each other.

[0014] The calculation unit 12 includes a vegetation index calculation unit 120 , a vegetation index information storage unit 121 , and a data extension unit 122 . The vegetation index calculation unit 120 calculates the vegetation index from image data obtained by photographing the farm field. The acquired image data is data measured by a sensor from above the field. For example, each acquired image data is image data obtained by photographing an area including at least the field from above. This photographing may be performed by a person using a sensor, or may be performed by a sensor installed or mounted on a structure such as a steel tower in or near the field, a drone, an airplane, or a satellite. The field photographed here is the same field as the field from which field survey data is obtained, and the same plants are growing there.

[0015] The acquired image data is, for example, data in which the luminance value of each pixel is expressed by RGB values ​​(red, green, blue values), near-infrared light values ​​(NIR values), and red edge light values ​​(Red Edge values). More specifically, the acquired image data is data that includes latitude and longitude, and the luminance value of red light, the luminance value of blue light, the luminance value of green light, the luminance value of near-infrared light, and the luminance value of red edge light, which is the edge of red. In this case, the acquired image data r is r(lat,lon)=[Red(lat,lon),Blue(lat,lon),Green(lat,lon)],… NIR((lat,lon),RedEdge(lat,lon)) Here, (lat,lon) corresponds to (latitude, longitude) and indicates location information.

[0016] In this embodiment, the acquired image data may be any data that can be correlated with the field survey data. Furthermore, the date on which the acquired image data was measured may be different from the date on which the field survey was conducted to obtain the field survey data. This is because the correlation between the acquired image data and the field survey data is sufficient, and even if the dates are different, the tendency of variation in the growth conditions of plants in the field is unlikely to change significantly to the extent that it would affect the estimation of the number of stems.

[0017] The vegetation index calculation unit 120 performs processing to obtain a vegetation index from the acquired image data and outputs the processing result.

[0018] There are various types of vegetation indices, such as NDVI (Normalized Difference Vegetation Index), NDRE (Normalized Red Edge Index), and NDWI (Normalized Water Index). NDVI is a normalized value based on the difference in reflectance between near-infrared light and red light. Plants reflect near-infrared wavelengths, but absorb red wavelengths, which are necessary for photosynthesis. Therefore, by obtaining NDVI based on the difference in reflectance between near-infrared light and red light obtained by photographing plants, the higher this value, the greater the area of ​​the plant's leaves and stems and the amount of chlorophyll they contain. NDRE is a value obtained by emphasizing the relative intensity difference between red edge wavelength light and near-infrared light. Like NDVI, NDRE is a vegetation index that correlates with the amount of chlorophyll in leaves, but compared to NDVI, it is used as an indicator of plant health in the later stages of growth and to detect plant growth problems earlier than NDVI. There are two types of NDWI: one that defines the amount of water contained in vegetation, and one that defines the amount of water contained in the earth's surface. Of these, the NDWI that defines the amount of water contained in vegetation is a value normalized based on the difference in reflectance between near-infrared light and short-wavelength infrared light, and is an index related to the amount of water contained in plants. In addition, the NDWI that defines the amount of water contained in vegetation may also be a value normalized based on the difference in reflectance between green light and short-wavelength infrared light. NDWI, which defines the amount of water contained in the earth's surface, is a normalized value based on the difference in reflectance between red light and short-wavelength infrared light, and is an index of the amount of water contained in the soil. NDWI, which defines the amount of water contained in the earth's surface, may also be normalized based on the difference in reflectance between green light and near-infrared light.

[0019] Here, during the plant growth period, at one time, among several vegetation indices, NDVI may have a particularly strong correlation with field survey data, at another time, NDRE may have a particularly strong correlation with field survey data, and at another time, there may be a strong correlation between both NDVI and NDRE and field survey data, and so on. In this way, the vegetation indices correlated with field survey data may vary depending on the time period.

[0020] However, even if NDVI has a particularly strong correlation with field survey data, it is not necessarily the vegetation index that can estimate stalk number with the highest accuracy. A vegetation index that slightly modifies NDVI may be able to estimate stalk number with even higher accuracy. For example, a vegetation index that uses the difference in red and green reflectance instead of the difference in near-infrared and red reflectance in NDVI may be able to estimate stalk number with higher accuracy. A vegetation index that weights near-infrared and red light in NDVI at a 1:2 ratio may be able to estimate stalk number with higher accuracy. Furthermore, because soil composition varies from field to field and plants grow under the influence of soil, it is best to assume that the vegetation index that can accurately estimate stalk number will vary from field to field.

[0021] To address this issue, in this embodiment, a calculation formula for calculating a vegetation index (e.g., NDVI) is expanded with data to generate a new calculation formula for calculating a vegetation index. Here, "expanding data" means generating a new calculation formula for calculating a vegetation index (e.g., NDVI) by modifying the calculation formula for the vegetation index. For example, the calculation formula for calculating a vegetation index can be modified by changing the combination of light components used in the calculation formula for the vegetation index or by multiplying the light components by a weighting coefficient. Then, by repeatedly conducting trials to confirm how accurately the number of stalks can be estimated when the new vegetation index is used, it is possible to identify a vegetation index that can estimate the number of stalks with the highest accuracy.

[0022] Specifically, the data extension unit 122 generates a calculation formula for calculating a new vegetation index by data extension of a calculation formula for calculating a vegetation index (for example, NDVI). The data extension unit 122 generates a calculation formula for calculating a new vegetation index by changing the combination of light components used in the calculation formula for the base vegetation index. For example, when the base vegetation index is NDVI, the calculation formula for calculating the vegetation index can be expressed by equation (1). In equation (1), I is the vegetation index. NIR is the light component of near-infrared light. Red is the light component of red light.

[0023] I=(NIR-Red) / (NIR+Red)...Equation (1)

[0024] In other words, NDVI is a vegetation index calculated based on a combination of near-infrared light and red light. The data expansion unit 122 changes the combination of light components in the NDVI calculation formula shown in formula (1) to generate a calculation formula for a new vegetation index, for example, a combination of blue light and green light. In this case, the calculation formula for the new vegetation index can be expressed as formula (2). I# in formula (2) is the new vegetation index. Blue is the light component of blue light. Green is the light component of green light.

[0025] I#=(Blue-Green) / (Blue+Green) …Equation (2)

[0026] Furthermore, the data expansion unit 122 changes the weighting coefficient by which the light components are multiplied in the NDVI calculation formula shown in formula (1), and sets the ratio of near-infrared light to red light to, for example, 1:2. In this case, the calculation formula for determining the new vegetation index can be expressed as formula (3). In formula (3), I# is the new vegetation index. NIR is the near-infrared light component. Red is the red light component.

[0027] I# = (NIR - 2 × Red) / (NIR + 2 × Red) ... Equation (3)

[0028] The data extension unit 122 performs data extension based on five vegetation indices, for example, NDVI, IRVI, RGBVI, VARI, and EVI, and generates a calculation formula for calculating a new vegetation index from each vegetation indices. The calculation formula for calculating each vegetation indices, IRVI, RGBVI, VARI, and EVI, can be expressed as follows:

[0029] IRVI=Blue×NIR / Red RGBVI={Green^2-(Red×Blue)} / {Green^2+(Red×Blue)} VARI=(Green-Red) / (Green+Red-Blue) EVI=(NIR-Red) / (NIR+6×Red-7.5×Blue+1)

[0030] The data expansion unit 122 stores the generated calculation formula (calculation formula for calculating a new vegetation index) in the vegetation index information storage unit 121. FIG. 2 is a diagram showing an example of information stored in the vegetation index information storage unit 121. The vegetation index information storage unit 121 stores information corresponding to each item, namely, a vegetation index ID, a base vegetation index, and a vegetation index calculation formula. The vegetation index ID indicates an identifier for identifying a new vegetation index. The base vegetation index indicates the vegetation index used as a basis for generating the calculation formula for the new vegetation index. The vegetation index calculation formula indicates the calculation formula for the new vegetation index. In this diagram, the new vegetation index identified by the vegetation index ID (I1) is a vegetation index calculated from a calculation formula generated based on NDVI, and shows that the combination of near-infrared light and red light in NDVI is replaced by a combination of blue light and green light. In addition, the new vegetation index identified by the vegetation index ID (I2) is a vegetation index calculated from a calculation formula generated based on NDVI, and is a combination of near-infrared light and red light in NDVI, i.e., it is NDVI itself. In this way, the new vegetation index may include the original vegetation index.

[0031] The vegetation index calculation unit 120 calculates the vegetation index from the acquired image data using a calculation formula stored in the vegetation index information storage unit 121. For example, the vegetation index calculation unit 120 may calculate NDVI and RGBVI for each latitude and longitude of the acquired image data. The IDs of the NDVI and RGBVI stored in the vegetation index information storage unit 121 shown in Figure 2 are assumed to be I2 and I130, respectively. At this time, n j is the jth vegetation index (j is any natural number) calculated from image data r, n2: NDVI calculated from image data r n 130 :RGBVI calculated from image data r The vegetation index for each latitude and longitude is expressed as follows: n2(lat,lon),n 130 (lat,lon) It can be expressed as follows.

[0032] In the above description, the field photographed to obtain the acquired image data is the same field as the field from which the field survey data is obtained. However, when calculating the correlation with the field survey data, the photographed field and the field from which the field survey data is obtained are the same field. On the other hand, when estimating the number of stems using the calculated correlation, i.e., when estimating the number of stems in the field from which the spray amount of the spray product is calculated, the field may be the same as the field from which the field survey data is obtained, or a different field. If the field is different, it is sufficient that the same plants are planted as in the field from which the field survey data was obtained. Furthermore, if the field is different, the accuracy of estimating the number of stems can be improved if the plants are the same as those in the field from which the field survey data was obtained and the field is located in a neighboring area of ​​the field from which the field survey data was obtained. This is because, even if the fields are different, the growth conditions of the plants are more similar in neighboring areas.

[0033] Furthermore, if the time when the acquired image data is measured and the time when the field survey is conducted correspond to the growth stage of the target for which the growth index value is to be estimated, the acquired image data and field survey data corresponding to the time of the target for which the growth index value is to be estimated can be obtained to generate the estimation model described below. In this case, the estimation accuracy of the growth index value can be improved.

[0034] Furthermore, although the vegetation index calculation unit 120 has been described as inputting a single acquired image data, it may also be configured to input multiple acquired image data acquired using the same or different methods and calculate the vegetation index from the acquired image data.

[0035] The filter processing unit 13 performs filtering on the vegetation index for each position in the field obtained by the vegetation index calculation unit 120. The purpose of the filtering is to remove information that is unnecessary for estimating the growth index value. For example, when constructing an estimation model for the number of stems, processing such as reducing the resolution of divided areas of the high-resolution acquired image data is performed to prevent unnecessarily emphasizing areas with a small or large number of stems locally. The filter used here may be a predetermined filter, or at least one of a median filter, a band-elimination filter, etc., may be used, or may be adaptively determined using a convolutional neural network that simultaneously estimates the optimal filter and the estimation model.

[0036] The field survey point data extraction unit 14 extracts a vegetation index corresponding to a position (divided area) included in the field survey data obtained by the field survey data acquisition unit 11 from the vegetation index for each position of the field after filtering, obtained by the filter processing unit 13. For example, the field survey point data extraction unit 14 cuts out, from the divided areas for which the vegetation index has been obtained by the filter processing unit 13, a divided area in which a field survey has been carried out. As a result, of the vegetation indices at each position in the field obtained by the vegetation index calculation unit 120 and the filter processing unit 13, the vegetation indices at the positions corresponding to the positions at which the field survey data was obtained are extracted.

[0037] The estimation model generation unit 15 generates an estimation model. The estimation model is a model that estimates the growth state of plants from the vegetation index. The estimation model generation unit 15 stores information such as the configuration of the generated estimation model in the estimation model information storage unit 19. FIG. 3 is a diagram for explaining the processing performed by the estimation model generation unit 15. As shown in this figure, the estimation model generation unit 15 generates a function indicating the relationship between the vegetation index and the growth index value (e.g., the number of stems) obtained from field survey data as an estimation model by applying regression analysis to the relationship between the vegetation index and the growth index value. The regression analysis may use linear regression, polynomial regression, Bayesian linear regression, Bayesian neural network, or the like. A learning model may be generated by learning the relationship between the number of stems and the vegetation index for each position.

[0038] The filter processing unit 16 performs filtering using a filter function similar to that of the filter processing unit 13 on the acquired image data, which is data output from the vegetation index calculation unit 120 and obtained by measuring the field for which the growth index value is to be estimated, i.e., the field for which the amount of sprayed material is to be estimated. Here, the filter function used in the filter processing unit 13 is passed to the filter processing unit 16, so that the filter processing unit 16 performs filtering on the vegetation index based on the acquired image for estimating the number of stems using a filter function similar to the filter function used to generate the estimation model. By performing such filtering, the filter processing unit 16 obtains n j By applying filter k to the feature vector x j , k (lat,lon) is obtained. x j , k is the vegetation index n after filtering by filter k j is.

[0039] The estimation accuracy calculation unit 17 calculates the estimation accuracy of the estimation model generated by the estimation model generation unit 15. The estimation accuracy calculation unit 17 calculates the estimation accuracy by comparing the estimation results based on the estimation model with the results of the field survey data. 4 is a diagram for explaining the processing performed by the estimation accuracy calculation unit 17. As shown in this figure, the estimation accuracy calculation unit 17 calculates the error between the stalk number estimation result obtained using the estimation model and the stalk number obtained using field survey data, and calculates statistical quantities of the calculated error, such as the mean value, maximum value, minimum value, mode, variance, and standard deviation, as the estimation accuracy. The estimation accuracy calculation unit 17 stores the calculated estimation accuracy in the estimation model information storage unit 19.

[0040] The estimation model suitability determination unit 18 determines the suitability of the estimation model generated using the new vegetation index. For example, the estimation model suitability determination unit 18 determines that an estimation model whose estimation accuracy is equal to or greater than a threshold is an appropriate estimation model. The estimation model suitability determination unit 18 may select the estimation model with the highest estimation accuracy and determine the selected estimation model to be an appropriate estimation model.

[0041] The estimation model information storage unit 19 stores information related to the estimation model. FIG. 5 is a diagram showing an example of information stored in the estimation model information storage unit 19. The estimation model information storage unit 19 stores information corresponding to each item, for example, an estimation model ID, a vegetation index ID, a model configuration, and estimation accuracy. The estimation model ID is an identifier for identifying the estimation model. The vegetation index ID is an identifier for the vegetation index used to generate the estimation model identified by the estimation model ID. The model configuration is the configuration of the model generated by the estimation model generation unit 15, such as a polynomial. The estimation accuracy is the estimation accuracy calculated by the estimation accuracy calculation unit 17.

[0042] The above-mentioned field survey data acquisition unit 11, calculation unit 12, filter processing unit 13, field survey point data extraction unit 14, estimation model generation unit 15, filter processing unit 16, estimation accuracy calculation unit 17, and estimation model suitability determination unit 18 may be configured as a processing device such as a CPU (Central Processing Unit) or a dedicated electronic circuit. Furthermore, the functions of the estimation model generation device 1 may be implemented as a single device by being installed in a single computer, or may be implemented as a system available on the cloud by being installed in a server device connected to a network such as the Internet. Furthermore, data acquired from outside, values ​​calculated in each unit, and data prepared in advance (e.g., reference data) may be stored in the storage unit. In this case, the storage unit and quick reference table storage unit 30 are configured with a storage medium such as a hard disk drive (HDD), flash memory, electrically erasable programmable read-only memory (EEPROM), random access read / write memory (RAM), read-only memory (ROM), or any combination of these storage media. Furthermore, these storage units may use, for example, non-volatile memory.

[0043] Next, the operation of the estimation model generating device 1 in the above-described embodiment will be described. FIG. 6 is a flowchart illustrating the flow of processing by the estimation model generating device 1 to generate an estimation model. To generate an estimation model, a field survey is conducted and the field is measured using sensors. The field survey data acquisition unit 11 acquires growth index values ​​(e.g., wheat stalk number and leaf color) based on the results of the field survey as field survey data. Furthermore, if it is assumed that nitrogen fertilizer will be sprayed during the flag leaf stage, which is one of the wheat growth stages, this field survey is preferably conducted during the flag leaf stage. The results of the field survey are acquired based on the content entered by an input person via an input device such as a keyboard, mouse, or panel (step S101). The vegetation index calculation unit 120 acquires the acquired image data (step S102). The estimation model generation device 1 initializes a variable n, for example, setting n=1 (step S103). The vegetation index calculation unit 120 calculates the nth vegetation index from the acquired image data using a calculation formula for the nth vegetation index (step S104). The nth vegetation index is the nth vegetation index among the new vegetation indices calculated by the vegetation index calculation unit 120. Here, the order of steps S101 and S102 may be interchanged. The field survey data and survey-acquired image data may be data obtained up to the year prior to the year in which the spraying amount is being considered.

[0044] Once the vegetation indices are calculated, the filter processing unit 13 performs a filter process on the n-th vegetation indices at each position in the field calculated by the vegetation index calculation unit 120 (step S105). The field survey point data extraction unit 14 extracts the n-th vegetation indices corresponding to each position (divided area) where the field survey was performed from the n-th vegetation indices after filtering by the filter processing unit 13 (step S106). The estimation model generation unit 15 generates an n-th estimation model that estimates the growth state of a plant (e.g., the number of stems) from the relationship between the field survey data (e.g., the number of stems) and the n-th vegetation index after filtering by the filtering unit 16 (step S107). The estimation accuracy calculation unit 17 calculates the estimation accuracy of the n-th estimation model (step S108). The estimation model generation device 1 increments the variable n (step S109). The estimation model generation device 1 determines whether the variable n exceeds the number of vegetation indices to be tried (step S110). The number of vegetation indices to be tried is the number of new vegetation indices used in the trial, that is, the process of confirming how accurately the number of stems can be estimated. If the variable n exceeds the number of vegetation indices to be tried, the estimation model suitability determination unit 18 determines the suitability of the estimation models based on the estimation accuracy of each estimation model. For example, the estimation model suitability determination unit 18 selects the model with the highest estimation accuracy from among the estimation models as the optimal model (step S111). On the other hand, in step S110, if the variable n does not exceed the number of vegetation indices to be tried, the estimation model generating device 1 returns to step S104 and repeatedly calculates the vegetation indices and generates an estimation model based on the calculated vegetation indices.

[0045] In the embodiment described above, the estimation model generating device 1 has been described as estimating the number of stalks for wheat, but it may also be configured to estimate the number of stalks for other plants, such as rice, barley, rye, etc.

[0046] Furthermore, in the above-described embodiment, a case has been exemplified in which multiple new vegetation indices are calculated and an estimation model is generated based on each of the calculated new vegetation indices, but this is not limiting. It is sufficient to generate at least one estimation model based on one new vegetation index obtained from a calculation formula generated by data-extending the calculation formula for the original vegetation index. If it is possible to compare the estimation accuracy when using the original vegetation index with the estimation accuracy when using the new vegetation index, it is possible to find an estimation model that will further improve the estimation accuracy.

[0047] According to the embodiment described above, a new calculation formula for calculating a vegetation index can be generated by modifying the calculation formula for calculating a vegetation index, and thus an estimation model based on the new vegetation index can be generated. Therefore, it is possible to generate an estimation model with higher estimation accuracy, and it becomes possible to estimate the growth state of plants in a field using a vegetation index that is suitable for estimating the growth state of plants.

[0048] Each unit of the estimation model generating device in the above-described embodiment may be implemented by a computer. In this case, a program for implementing the functions may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include devices that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or devices that store programs for a fixed period of time, such as volatile memory within a computer system that serves as a server or client. The program may be for implementing some of the above-described functions, or may be capable of implementing the above-described functions in combination with a program already stored in the computer system, or may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0049] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention. [Explanation of symbols]

[0050] 1...Estimation model generating device, 11...Field survey data acquisition unit, 12...Calculation unit, 13...Filter processing unit, 14...Field survey point data extraction unit, 15...Estimation model generating unit, 16...Filter processing unit, 17...Estimation accuracy calculation unit, 18...Estimation model suitability determination unit, 19...Estimation model information storage unit, 120...Vegetation index calculation unit, 121...Vegetation index information storage unit, 122...Data expansion unit

Claims

1. a data expansion unit that generates a new formula for calculating a vegetation index by modifying the formula for calculating a vegetation index; a vegetation index calculation unit that calculates a vegetation index from acquired image data obtained by photographing the field using a calculation formula for determining a new vegetation index generated by the data extension unit; an estimation model generation unit that generates an estimation model that estimates the growth state of plants at each position in the field based on the relationship between a growth index value based on an evaluation item that indicates the growth state of plants in the field and the vegetation index calculated by the vegetation index calculation unit; and an estimation model appropriateness determination unit that determines appropriateness of the estimation model based on the estimation accuracy of the estimation model generated by the estimation model generation unit; An estimation model generating device comprising:

2. the data expansion unit generates a plurality of calculation formulas for obtaining new vegetation indices from one calculation formula, the vegetation index calculation unit calculates a plurality of vegetation indices by applying a plurality of calculation formulas generated by the data extension unit to the acquired image data, the estimation model generation unit generates a plurality of estimation models corresponding to the vegetation indices calculated by the vegetation index calculation unit, the estimation model appropriateness determination unit selects an estimation model with high estimation accuracy from each of the plurality of estimation models generated by the estimation model generation unit. The estimation model generating device according to claim 1 .

3. the data extension unit generates a new calculation formula for the vegetation index by changing a combination of light components in the calculation formula for obtaining the vegetation index; The estimation model generating device according to claim 1 .

4. the data expansion unit generates a new calculation formula for the vegetation index by changing a weighting coefficient of a light component in the calculation formula for obtaining the vegetation index; The estimation model generating device according to claim 1 .

5. the estimation model generation unit generates an estimation model that estimates the growth index value as a growth state of the plant at each position in the field from the relationship between the growth index value, which includes at least one of the number of stems, the number of upper stems, plant height, water content, water content rate, and leaf color, and a vegetation index; The estimation model generating device according to claim 1 .

6. the estimation model generation unit generates an estimation model that estimates the growth index value as the growth state of plants at each position in the field from a relationship between a vegetation index and a calculated value calculated by applying a function to the growth index value measured based on an evaluation item indicating the growth state of plants in the field. The estimation model generating device according to claim 1 .

7. The data expansion unit generates a new formula for calculating the vegetation index by modifying the formula for calculating the vegetation index; a vegetation index calculation unit calculates a vegetation index from acquired image data obtained by photographing the field using a calculation formula for determining a new vegetation index generated by the data extension unit; an estimation model generation unit generates an estimation model that estimates the growth state of plants at each position in the field from the relationship between a growth index value based on an evaluation item indicating the growth state of plants in the field and the vegetation index calculated by the vegetation index calculation unit; and an estimation model appropriateness determination unit determines appropriateness of the estimation model based on the estimation accuracy of the estimation model generated by the estimation model generation unit; Estimation model generation method.

Images