WO2019064455A1 - Fertilizer applicator control system, fertilizer applicator control method, and program - Google Patents
Fertilizer applicator control system, fertilizer applicator control method, and program Download PDFInfo
- Publication number
- WO2019064455A1 WO2019064455A1 PCT/JP2017/035306 JP2017035306W WO2019064455A1 WO 2019064455 A1 WO2019064455 A1 WO 2019064455A1 JP 2017035306 W JP2017035306 W JP 2017035306W WO 2019064455 A1 WO2019064455 A1 WO 2019064455A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- growth
- crop
- image
- fertilizer
- drone
- Prior art date
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01C—PLANTING; SOWING; FERTILISING
- A01C15/00—Fertiliser distributors
Definitions
- the present invention relates to a fertilizer application control system, a fertilizer application control method, and a program for controlling a fertilizer application to fertilize poorly grown crops.
- Fertilizer technology has evolved in recent years. For example, a seedling transplanting machine having an appropriate fertilization amount according to the state of a field has been provided (Patent Document 1).
- the present invention analyzes the image taken by the drone camera to determine the growth defect of the crop, and applies the fertilization machine to pinpoint the crop determined to be the growth defect. It is an object of the present invention to provide a fertilizer application control system, a fertilizer application control method and a program to be controlled.
- the present invention provides the following solutions.
- the invention according to the first aspect is a fertilizer control system for controlling a fertilizer applicator to fertilize crops with poor growth, which comprises: an image acquisition unit for acquiring an image captured by a drone camera; A poor growth determining means for analyzing and determining poor growth of agricultural products, a position estimating means for estimating the position of agricultural products determined to be poor growth, and a fertilizing machine for fertilizing agricultural products at the estimated position And a fertilizer control system for controlling the fertilizer application.
- the invention according to the first aspect is a fertilizer application control method for controlling a fertilizer application to fertilize crops with poor growth, which comprises an image acquisition step of acquiring an image captured by a drone camera, and the image acquisition step.
- the invention according to the first aspect includes a computer, an image acquisition step of acquiring an image captured by a drone camera, a growth failure determination step of analyzing the image to determine growth failure of a crop, and the growth
- a program is provided for performing a position estimation step of estimating the position of an agricultural product determined to be defective and a fertilizer application control step of controlling a fertilizer application device to fertilize the agricultural product at the estimated position.
- FIG. 1 is a schematic view of a fertilizer application control system.
- FIG. 2 shows an example of a crop determined to be unhealthy.
- the fertilizer application control system of the present invention is a system for controlling a fertilizer application to fertilize poorly grown crops.
- FIG. 1 is a schematic view of a fertilizer application control system according to a preferred embodiment of the present invention.
- the fertilizing machine control system includes an image acquiring unit, a growth defect judging unit, a position estimating unit, and a fertilizing machine control unit, which are realized by the control unit reading a predetermined program.
- drone control means may be provided similarly. These may be application based, cloud based or otherwise. Each means described above may be realized by a single computer or may be realized by two or more computers (for example, in the case of a server and a terminal).
- An image acquisition means acquires the image imaged with the camera of the drone.
- the image may be a moving image or a still image.
- the camera may be any camera provided in the drone, either a digital camera or a camera of a smartphone. Real-time images are preferable for real-time fertilization.
- the poor growth determination means analyzes the image to determine the poor growth of the crop.
- Machine learning may improve the accuracy of image analysis. For example, machine learning is performed using past images as teacher data. For example, as shown in FIG. 2, image analysis is performed to determine the growth defect of a crop.
- the poor growth determination means may analyze the image and determine that the crop is poor growth if the NDVI (Normalized Difference Vegetation Index) satisfies a predetermined condition.
- NDVI is an index showing the distribution status and activity of vegetation (normalized difference vegetation index).
- R is the reflectance of red in the visible range of the data
- IR is the reflectance in the near infrared range of the data. That is, since NDVI indicates a normalized value between -1 and 1, the larger the positive number is, the thicker the vegetation is. Therefore, if the image is analyzed and the NDVI is known, the growth defect can be determined.
- the poor growth determination means may analyze the image and determine that the crop is poor growth when the color satisfies a predetermined condition. It differs from NDVI in that it analyzes the color of the visible region. If the color of the normal growth state of cabbage is machine-learned and a color different from that is detected by image analysis, it may be determined as poor growth. For example, if various yellow-green cabbage images that are normal colors of cabbage are machine-learned, it is possible to determine poor growth if brown is detected by image analysis.
- the position estimation means estimates the position of the crop determined to be nonviable. From the GPS equipped with the drone, the shooting height with the drone, the shooting angle, and the shooting direction, it is possible to deduce the position of the crop determined to be unhealthy. For example, in the case of the shooting height H and the shooting angle ⁇ , the position of the crop determined to be poor in growth is the latitude / longitude obtained by adding Htan ⁇ to the latitude / longitude of GPS in consideration of the shooting direction.
- the fertilization machine control means controls the fertilization machine to fertilize the crop at the estimated position.
- the tractor may be controlled to fertilize the crop at the inferred position.
- the fertilization dedicated drone may be controlled to fertilize the crop at the estimated position.
- Fertilizer-specific drones are drones that have been modified to carry heavy fertilizers. By selectively using the imaging dedicated drone and the fertilization dedicated drone, efficient fertilization becomes possible.
- the drone control means controls the drone to fly again and image again at the estimated position. For example, since there is no possibility that the determination of the growth defect is an incorrect determination, the drone is re-flyed and photographed again to care for the incorrect determination. By analyzing the image captured again, it is possible to double check the determination of growth failure. [Description of operation]
- the fertilizing machine control method of the present invention is a method of controlling a fertilizing machine to fertilize poorly grown crops.
- the fertilization machine control method includes an image acquisition step, a growth defect determination step, a position estimation step, and a fertilization machine control step. Also, a drone control step may be provided.
- the image acquisition step acquires an image captured by the drone camera.
- the image may be a moving image or a still image.
- the camera may be any camera provided in the drone, either a digital camera or a camera of a smartphone. Real-time images are preferable for real-time fertilization.
- the poor growth determination step analyzes the image to determine the poor growth of the crop.
- Machine learning may improve the accuracy of image analysis. For example, machine learning is performed using past images as teacher data. For example, as shown in FIG. 2, image analysis is performed to determine the growth defect of a crop.
- the poor growth determination step may analyze the image and determine that the crop is poor growth when the NDVI (Normalized Difference Vegetation Index) satisfies a predetermined condition.
- NDVI is an index showing the distribution status and activity of vegetation (normalized difference vegetation index).
- R is the reflectance of red in the visible range of the data
- IR is the reflectance in the near infrared range of the data. That is, since NDVI indicates a normalized value between -1 and 1, the larger the positive number is, the thicker the vegetation is. Therefore, if the image is analyzed and the NDVI is known, the growth defect can be determined.
- the poor growth determination step may determine that the crop is poor growth when the image is analyzed and the color satisfies a predetermined condition. It differs from NDVI in that it analyzes the color of the visible region. If the color of the normal growth state of cabbage is machine-learned and a color different from that is detected by image analysis, it may be determined as poor growth. For example, if various yellow-green cabbage images that are normal colors of cabbage are machine-learned, it is possible to determine poor growth if brown is detected by image analysis.
- the position estimation step estimates the position of the crop determined to be nonviable. From the GPS equipped with the drone, the shooting height with the drone, the shooting angle, and the shooting direction, it is possible to deduce the position of the crop determined to be unhealthy. For example, in the case of the shooting height H and the shooting angle ⁇ , the position of the crop determined to be poor in growth is the latitude / longitude obtained by adding Htan ⁇ to the latitude / longitude of GPS in consideration of the shooting direction.
- the fertilizing machine control step controls the fertilizing machine to fertilize the crop at the inferred position.
- the tractor may be controlled to fertilize the crop at the inferred position.
- the fertilization dedicated drone may be controlled to fertilize the crop at the estimated position.
- Fertilizer-specific drones are drones that have been modified to carry heavy fertilizers. By selectively using the imaging dedicated drone and the fertilization dedicated drone, efficient fertilization becomes possible.
- the drone control step controls the drone to fly again and image again at the estimated position. For example, since there is no possibility that the determination of the growth defect is an incorrect determination, the drone is re-flyed and photographed again to care for the incorrect determination. By analyzing the image captured again, it is possible to double check the determination of growth failure.
- the above-described means and functions are realized by a computer (including a CPU, an information processing device, and various terminals) reading and executing a predetermined program.
- the program may be, for example, an application installed on a computer, or a SaaS (software as a service) provided from a computer via a network, for example, a flexible disk, a CD It may be provided in the form of being recorded in a computer readable recording medium such as a CD-ROM or the like, a DVD (DVD-ROM, DVD-RAM or the like).
- the computer reads the program from the recording medium, transfers the program to the internal storage device or the external storage device, stores it, and executes it.
- the program may be recorded in advance in a storage device (recording medium) such as, for example, a magnetic disk, an optical disk, or a magneto-optical disk, and may be provided from the storage device to the computer via a communication line.
- nearest neighbor method naive Bayes method
- decision tree naive Bayes method
- support vector machine e.g., support vector machine
- reinforcement learning e.g., reinforcement learning, etc.
- deep learning may be used in which feature quantities for learning are generated by using a neural network.
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Soil Sciences (AREA)
- Environmental Sciences (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
[Problem] To analyze images captured by a camera of a drone to determine growth failure of crops, and control a fertilizer applicator such that the application of fertilizer is pinpointed to crops for which growth failure has been determined. [Solution] This fertilizer applicator control system, which controls a fertilizer applicator to cause the fertilizer applicator to apply fertilizer to crops which have failed to grow is provided with: an image acquisition means for acquiring images captured by a camera of a drone; a growth failure determination means which analyzes the images to determine growth failure of crops; a position estimation means for estimating the positions of the crops for which growth failure has been determined; and a fertilizer applicator control means for controlling the fertilizer applicator such that fertilizer is applied to the crops in the estimated positions.
Description
本発明は、施肥機を制御して生育不良の農作物に施肥させる施肥機制御システム、施肥機制御方法およびプログラムに関する。
The present invention relates to a fertilizer application control system, a fertilizer application control method, and a program for controlling a fertilizer application to fertilize poorly grown crops.
近年、施肥の技術が進化している。例えば、圃場の状態に応じて施肥量が適切な苗移植機が提供されている（特許文献１）。
Fertilizer technology has evolved in recent years. For example, a seedling transplanting machine having an appropriate fertilization amount according to the state of a field has been provided (Patent Document 1).
しかしながら、特許文献１の装置では、生育不良の農作物に対してピンポイントに施肥出来ない問題がある。
However, in the device of Patent Document 1, there is a problem that the pinpoint can not be fertilized with respect to the crop with poor growth.
本発明は、上記課題に鑑み、ドローンのカメラで撮像された画像を解析して、農作物の生育不良を判定し、生育不良と判定された農作物に対してピンポイントに施肥するように施肥機を制御する施肥機制御システム、施肥機制御方法およびプログラムを提供することを目的とする。
In view of the above problems, the present invention analyzes the image taken by the drone camera to determine the growth defect of the crop, and applies the fertilization machine to pinpoint the crop determined to be the growth defect. It is an object of the present invention to provide a fertilizer application control system, a fertilizer application control method and a program to be controlled.
本発明では、以下のような解決手段を提供する。
The present invention provides the following solutions.
第１の特徴に係る発明は、施肥機を制御して、生育不良の農作物に施肥させる施肥機制御システムであって、ドローンのカメラで撮像された画像を取得する画像取得手段と、前記画像を解析して、農作物の生育不良を判定する生育不良判定手段と、前記生育不良と判定された農作物の位置を推測する位置推測手段と、前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御手段と、を備える施肥機制御システムを提供する。
The invention according to the first aspect is a fertilizer control system for controlling a fertilizer applicator to fertilize crops with poor growth, which comprises: an image acquisition unit for acquiring an image captured by a drone camera; A poor growth determining means for analyzing and determining poor growth of agricultural products, a position estimating means for estimating the position of agricultural products determined to be poor growth, and a fertilizing machine for fertilizing agricultural products at the estimated position And a fertilizer control system for controlling the fertilizer application.
第１の特徴に係る発明は、施肥機を制御して、生育不良の農作物に施肥させる施肥機制御方法であって、ドローンのカメラで撮像された画像を取得する画像取得ステップと、前記画像を解析して、農作物の生育不良を判定する生育不良判定ステップと、前記生育不良と判定された農作物の位置を推測する位置推測ステップと、前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御ステップと、を備える施肥機制御方法を提供する。
The invention according to the first aspect is a fertilizer application control method for controlling a fertilizer application to fertilize crops with poor growth, which comprises an image acquisition step of acquiring an image captured by a drone camera, and the image acquisition step. A poor growth determination step of analyzing and determining poor growth of agricultural products, a position estimation step of estimating the position of agricultural products determined to be poor growth, and a fertilizing machine so as to fertilize agricultural products at the estimated position Providing a fertilization machine control step of controlling the fertilization machine control step.
第１の特徴に係る発明は、コンピュータに、ドローンのカメラで撮像された画像を取得する画像取得ステップと、前記画像を解析して、農作物の生育不良を判定する生育不良判定ステップと、前記生育不良と判定された農作物の位置を推測する位置推測ステップと、前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御ステップと、をさせるためのプログラムを提供する。
The invention according to the first aspect includes a computer, an image acquisition step of acquiring an image captured by a drone camera, a growth failure determination step of analyzing the image to determine growth failure of a crop, and the growth A program is provided for performing a position estimation step of estimating the position of an agricultural product determined to be defective and a fertilizer application control step of controlling a fertilizer application device to fertilize the agricultural product at the estimated position.
生育不良の農作物に対してピンポイントに施肥できる。
It can be applied to pinpoint crops with poor growth.
以下、本発明を実施するための最良の形態について説明する。なお、これはあくまでも一例であって、本発明の技術的範囲はこれに限られるものではない。
The best mode for carrying out the present invention will be described below. This is merely an example, and the technical scope of the present invention is not limited to this.
本発明の施肥機制御システムは、施肥機を制御して生育不良の農作物に施肥させるシステムである。
The fertilizer application control system of the present invention is a system for controlling a fertilizer application to fertilize poorly grown crops.
本発明の好適な実施形態の概要について、図１に基づいて説明する。図１は、本発明の好適な実施形態である施肥機制御システムの概要図である。
An outline of a preferred embodiment of the present invention will be described based on FIG. FIG. 1 is a schematic view of a fertilizer application control system according to a preferred embodiment of the present invention.
図１にあるように、施肥機制御システムは、制御部が所定のプログラムを読み込むことで実現される、画像取得手段、生育不良判定手段、位置推測手段、施肥機制御手段を備える。また図示しないが、同様に、ドローン制御手段を備えてもよい。これらは、アプリケーション型、クラウド型またはその他であってもよい。上述の各手段が、単独のコンピュータで実現されてもよいし、２台以上のコンピュータ（例えば、サーバと端末のような場合）で実現されてもよい。
As shown in FIG. 1, the fertilizing machine control system includes an image acquiring unit, a growth defect judging unit, a position estimating unit, and a fertilizing machine control unit, which are realized by the control unit reading a predetermined program. Also, although not shown, drone control means may be provided similarly. These may be application based, cloud based or otherwise. Each means described above may be realized by a single computer or may be realized by two or more computers (for example, in the case of a server and a terminal).
画像取得手段は、ドローンのカメラで撮像された画像を取得する。画像は動画でも静止画でもよい。カメラは、ドローンに備えられたカメラであれば、デジタルカメラであっても、スマートフォンのカメラであっても、何でもよい。リアルタイムに施肥させるにはリアルタイムな画像の方が好ましい。
An image acquisition means acquires the image imaged with the camera of the drone. The image may be a moving image or a still image. The camera may be any camera provided in the drone, either a digital camera or a camera of a smartphone. Real-time images are preferable for real-time fertilization.
生育不良判定手段は、画像を解析して農作物の生育不良を判定する。機械学習によって画像解析の精度を向上させてもよい。例えば、過去画像を教師データとして機械学習を行う。例えば図２のように、画像解析して農作物の生育不良を判定する。
The poor growth determination means analyzes the image to determine the poor growth of the crop. Machine learning may improve the accuracy of image analysis. For example, machine learning is performed using past images as teacher data. For example, as shown in FIG. 2, image analysis is performed to determine the growth defect of a crop.
また、生育不良判定手段は、画像を解析してＮＤＶＩ（Ｎｏｒｍａｌｉｚｅｄ Ｄｉｆｆｅｒｅｎｃｅ Ｖｅｇｅｔａｔｉｏｎ Ｉｎｄｅｘ）が所定の条件を満たす場合に、農作物が生育不良であると判定してもよい。ＮＤＶＩとは、植生の分布状況や活性度を示す指標である（正規化差植生指数）。ＮＤＶＩは、 で与えられる。Ｒはデータの可視域赤の反射率、ＩＲはデータの近赤外域の反射率。つまり、ＮＤＶＩは－１から１の間に正規化した数値を示すので、正の大きい数字になるほど植生が濃いことを表す。従って、画像を解析してＮＤＶＩが分かれば、生育不良を判定することができる。
In addition, the poor growth determination means may analyze the image and determine that the crop is poor growth if the NDVI (Normalized Difference Vegetation Index) satisfies a predetermined condition. NDVI is an index showing the distribution status and activity of vegetation (normalized difference vegetation index). NDVI is given by R is the reflectance of red in the visible range of the data, and IR is the reflectance in the near infrared range of the data. That is, since NDVI indicates a normalized value between -1 and 1, the larger the positive number is, the thicker the vegetation is. Therefore, if the image is analyzed and the NDVI is known, the growth defect can be determined.
また、生育不良判定手段は、画像を解析して色が所定の条件を満たす場合に、農作物が生育不良であると判定してもよい。可視領域の色について解析を行う点で、ＮＤＶＩとは異なる。キャベツの正常な生育状態の色を機械学習しておき、画像解析によってそれとは異なる色を検出した場合に、生育不良と判定してもよい。例えば、キャベツの正常な色である黄緑色の様々なキャベツ画像を機械学習しておけば、画像解析によって茶色を検出した場合に、生育不良を判定できる。
Further, the poor growth determination means may analyze the image and determine that the crop is poor growth when the color satisfies a predetermined condition. It differs from NDVI in that it analyzes the color of the visible region. If the color of the normal growth state of cabbage is machine-learned and a color different from that is detected by image analysis, it may be determined as poor growth. For example, if various yellow-green cabbage images that are normal colors of cabbage are machine-learned, it is possible to determine poor growth if brown is detected by image analysis.
位置推測手段は、生育不良と判定された農作物の位置を推測する。ドローンに備えられたＧＰＳや、ドローンでの撮影高度、撮影角度、撮影向き、から生育不良と判定された農作物の位置を推測できる。例えば、撮影高度Ｈ、撮影角度θの場合、育成不良を判定された農作物の位置は、ＧＰＳの緯度経度に撮影向きを考慮してＨｔａｎθを加味した緯度経度となる。
The position estimation means estimates the position of the crop determined to be nonviable. From the GPS equipped with the drone, the shooting height with the drone, the shooting angle, and the shooting direction, it is possible to deduce the position of the crop determined to be unhealthy. For example, in the case of the shooting height H and the shooting angle θ, the position of the crop determined to be poor in growth is the latitude / longitude obtained by adding Htanθ to the latitude / longitude of GPS in consideration of the shooting direction.
施肥機制御手段は、推測された位置にある農作物に施肥させるように施肥機を制御する。例えば、推測された位置にある農作物に施肥させるようにトラクターを制御してもよい。例えば、推測された位置にある農作物に施肥させるように施肥専用ドローンを制御してもよい。施肥専用ドローンは、重い肥料を運べるように改良されたドローンである。撮像専用ドローンと施肥専用ドローンとを使い分けることで、効率的な施肥が可能となる。
The fertilization machine control means controls the fertilization machine to fertilize the crop at the estimated position. For example, the tractor may be controlled to fertilize the crop at the inferred position. For example, the fertilization dedicated drone may be controlled to fertilize the crop at the estimated position. Fertilizer-specific drones are drones that have been modified to carry heavy fertilizers. By selectively using the imaging dedicated drone and the fertilization dedicated drone, efficient fertilization becomes possible.
ドローン制御手段は、推測された位置に、ドローンを再度飛行させて再度撮像するように制御する。例えば、生育不良の判定が誤判定である可能性も無くはないので、誤判定をケアするためにも、ドローンを再度飛行させて再度撮影させる。再度撮像させた画像を解析することで、生育不良の判定の二重チェックが可能となる。
[動作の説明] The drone control means controls the drone to fly again and image again at the estimated position. For example, since there is no possibility that the determination of the growth defect is an incorrect determination, the drone is re-flyed and photographed again to care for the incorrect determination. By analyzing the image captured again, it is possible to double check the determination of growth failure.
[Description of operation]
[動作の説明] The drone control means controls the drone to fly again and image again at the estimated position. For example, since there is no possibility that the determination of the growth defect is an incorrect determination, the drone is re-flyed and photographed again to care for the incorrect determination. By analyzing the image captured again, it is possible to double check the determination of growth failure.
[Description of operation]
次に、施肥機制御方法について説明する。本発明の施肥機制御方法は、施肥機を制御して生育不良の農作物に施肥させる方法である。
Next, the fertilizer application control method will be described. The fertilizing machine control method of the present invention is a method of controlling a fertilizing machine to fertilize poorly grown crops.
施肥機制御方法は、画像取得ステップ、生育不良判定ステップ、位置推測ステップ、施肥機制御ステップを備える。また、ドローン制御ステップを備えてもよい。
The fertilization machine control method includes an image acquisition step, a growth defect determination step, a position estimation step, and a fertilization machine control step. Also, a drone control step may be provided.
画像取得ステップは、ドローンのカメラで撮像された画像を取得する。画像は動画でも静止画でもよい。カメラは、ドローンに備えられたカメラであれば、デジタルカメラであっても、スマートフォンのカメラであっても、何でもよい。リアルタイムに施肥させるにはリアルタイムな画像の方が好ましい。
The image acquisition step acquires an image captured by the drone camera. The image may be a moving image or a still image. The camera may be any camera provided in the drone, either a digital camera or a camera of a smartphone. Real-time images are preferable for real-time fertilization.
生育不良判定ステップは、画像を解析して農作物の生育不良を判定する。機械学習によって画像解析の精度を向上させてもよい。例えば、過去画像を教師データとして機械学習を行う。例えば図２のように、画像解析して農作物の生育不良を判定する。
The poor growth determination step analyzes the image to determine the poor growth of the crop. Machine learning may improve the accuracy of image analysis. For example, machine learning is performed using past images as teacher data. For example, as shown in FIG. 2, image analysis is performed to determine the growth defect of a crop.
また、生育不良判定ステップは、画像を解析してＮＤＶＩ（Ｎｏｒｍａｌｉｚｅｄ Ｄｉｆｆｅｒｅｎｃｅ Ｖｅｇｅｔａｔｉｏｎ Ｉｎｄｅｘ）が所定の条件を満たす場合に、農作物が生育不良であると判定してもよい。ＮＤＶＩとは、植生の分布状況や活性度を示す指標である（正規化差植生指数）。ＮＤＶＩは、 で与えられる。Ｒはデータの可視域赤の反射率、ＩＲはデータの近赤外域の反射率。つまり、ＮＤＶＩは－１から１の間に正規化した数値を示すので、正の大きい数字になるほど植生が濃いことを表す。従って、画像を解析してＮＤＶＩが分かれば、生育不良を判定することができる。
Further, the poor growth determination step may analyze the image and determine that the crop is poor growth when the NDVI (Normalized Difference Vegetation Index) satisfies a predetermined condition. NDVI is an index showing the distribution status and activity of vegetation (normalized difference vegetation index). NDVI is given by R is the reflectance of red in the visible range of the data, and IR is the reflectance in the near infrared range of the data. That is, since NDVI indicates a normalized value between -1 and 1, the larger the positive number is, the thicker the vegetation is. Therefore, if the image is analyzed and the NDVI is known, the growth defect can be determined.
また、生育不良判定ステップは、画像を解析して色が所定の条件を満たす場合に、農作物が生育不良であると判定してもよい。可視領域の色について解析を行う点で、ＮＤＶＩとは異なる。キャベツの正常な生育状態の色を機械学習しておき、画像解析によってそれとは異なる色を検出した場合に、生育不良と判定してもよい。例えば、キャベツの正常な色である黄緑色の様々なキャベツ画像を機械学習しておけば、画像解析によって茶色を検出した場合に、生育不良を判定できる。
Further, the poor growth determination step may determine that the crop is poor growth when the image is analyzed and the color satisfies a predetermined condition. It differs from NDVI in that it analyzes the color of the visible region. If the color of the normal growth state of cabbage is machine-learned and a color different from that is detected by image analysis, it may be determined as poor growth. For example, if various yellow-green cabbage images that are normal colors of cabbage are machine-learned, it is possible to determine poor growth if brown is detected by image analysis.
位置推測ステップは、生育不良と判定された農作物の位置を推測する。ドローンに備えられたＧＰＳや、ドローンでの撮影高度、撮影角度、撮影向き、から生育不良と判定された農作物の位置を推測できる。例えば、撮影高度Ｈ、撮影角度θの場合、育成不良を判定された農作物の位置は、ＧＰＳの緯度経度に撮影向きを考慮してＨｔａｎθを加味した緯度経度となる。
The position estimation step estimates the position of the crop determined to be nonviable. From the GPS equipped with the drone, the shooting height with the drone, the shooting angle, and the shooting direction, it is possible to deduce the position of the crop determined to be unhealthy. For example, in the case of the shooting height H and the shooting angle θ, the position of the crop determined to be poor in growth is the latitude / longitude obtained by adding Htanθ to the latitude / longitude of GPS in consideration of the shooting direction.
施肥機制御ステップは、推測された位置にある農作物に施肥させるように施肥機を制御する。例えば、推測された位置にある農作物に施肥させるようにトラクターを制御してもよい。例えば、推測された位置にある農作物に施肥させるように施肥専用ドローンを制御してもよい。施肥専用ドローンは、重い肥料を運べるように改良されたドローンである。撮像専用ドローンと施肥専用ドローンとを使い分けることで、効率的な施肥が可能となる。
The fertilizing machine control step controls the fertilizing machine to fertilize the crop at the inferred position. For example, the tractor may be controlled to fertilize the crop at the inferred position. For example, the fertilization dedicated drone may be controlled to fertilize the crop at the estimated position. Fertilizer-specific drones are drones that have been modified to carry heavy fertilizers. By selectively using the imaging dedicated drone and the fertilization dedicated drone, efficient fertilization becomes possible.
ドローン制御ステップは、推測された位置に、ドローンを再度飛行させて再度撮像するように制御する。例えば、生育不良の判定が誤判定である可能性も無くはないので、誤判定をケアするためにも、ドローンを再度飛行させて再度撮影させる。再度撮像させた画像を解析することで、生育不良の判定の二重チェックが可能となる。
The drone control step controls the drone to fly again and image again at the estimated position. For example, since there is no possibility that the determination of the growth defect is an incorrect determination, the drone is re-flyed and photographed again to care for the incorrect determination. By analyzing the image captured again, it is possible to double check the determination of growth failure.
上述した手段、機能は、コンピュータ（ＣＰＵ、情報処理装置、各種端末を含む）が、所定のプログラムを読み込んで、実行することによって実現される。プログラムは、例えば、コンピュータにインストールされるアプリケーションであってもよいし、コンピュータからネットワーク経由で提供されるＳａａＳ（ソフトウェア・アズ・ア・サービス）形態であってもよいし、例えば、フレキシブルディスク、ＣＤ（ＣＤ－ＲＯＭなど）、ＤＶＤ（ＤＶＤ－ＲＯＭ、ＤＶＤ－ＲＡＭなど）等のコンピュータ読取可能な記録媒体に記録された形態で提供されてもよい。この場合、コンピュータはその記録媒体からプログラムを読み取って内部記憶装置または外部記憶装置に転送し記憶して実行する。また、そのプログラムを、例えば、磁気ディスク、光ディスク、光磁気ディスク等の記憶装置（記録媒体）に予め記録しておき、その記憶装置から通信回線を介してコンピュータに提供するようにしてもよい。
The above-described means and functions are realized by a computer (including a CPU, an information processing device, and various terminals) reading and executing a predetermined program. The program may be, for example, an application installed on a computer, or a SaaS (software as a service) provided from a computer via a network, for example, a flexible disk, a CD It may be provided in the form of being recorded in a computer readable recording medium such as a CD-ROM or the like, a DVD (DVD-ROM, DVD-RAM or the like). In this case, the computer reads the program from the recording medium, transfers the program to the internal storage device or the external storage device, stores it, and executes it. Alternatively, the program may be recorded in advance in a storage device (recording medium) such as, for example, a magnetic disk, an optical disk, or a magneto-optical disk, and may be provided from the storage device to the computer via a communication line.
上述した機械学習の具体的なアルゴリズムとしては、最近傍法、ナイーブベイズ法、決定木、サポートベクターマシン、強化学習などを利用してよい。また、ニューラルネットワークを利用して、学習するための特徴量を自ら生成する深層学習（ディープラーニング）であってもよい。
As a specific algorithm of the above-mentioned machine learning, nearest neighbor method, naive Bayes method, decision tree, support vector machine, reinforcement learning, etc. may be used. In addition, deep learning may be used in which feature quantities for learning are generated by using a neural network.
以上、本発明の実施形態について説明したが、本発明は上述したこれらの実施形態に限るものではない。また、本発明の実施形態に記載された効果は、本発明から生じる最も好適な効果を列挙したに過ぎず、本発明による効果は、本発明の実施形態に記載されたものに限定されるものではない。
As mentioned above, although embodiment of this invention was described, this invention is not limited to these embodiment mentioned above. Further, the effects described in the embodiments of the present invention only list the most preferable effects resulting from the present invention, and the effects according to the present invention are limited to those described in the embodiments of the present invention is not.
As mentioned above, although embodiment of this invention was described, this invention is not limited to these embodiment mentioned above. Further, the effects described in the embodiments of the present invention only list the most preferable effects resulting from the present invention, and the effects according to the present invention are limited to those described in the embodiments of the present invention is not.
Claims (8)
- 施肥機を制御して、生育不良の農作物に施肥させる施肥機制御システムであって、
ドローンのカメラで撮像された画像を取得する画像取得手段と、
前記画像を解析して、農作物の生育不良を判定する生育不良判定手段と、
前記生育不良と判定された農作物の位置を推測する位置推測手段と、
前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御手段と、を備える施肥機制御システム。 A fertilizer control system that controls fertilization machines to fertilize poorly grown crops.
An image acquisition unit that acquires an image captured by the drone camera;
Growth defect determination means for analyzing the image to determine growth defects of crops;
Position estimating means for estimating the position of the crop determined to be the above-mentioned poor growth;
Fertilizer control system, comprising: fertilizer applicator control means for controlling the fertilizer applicator so as to fertilize the crop at the position estimated above. - 前記生育不良判定手段は、前記画像を解析して、ＮＤＶＩが所定の条件を満たす場合に、農作物が生育不良であると判定する
請求項１に記載の施肥機制御システム。 The fertilization machine control system according to claim 1, wherein the growth defect determination means determines that the crop is growth defect when the image is analyzed and the NDVI satisfies a predetermined condition. - 前記生育不良判定手段は、前記画像を解析して、色が所定の条件を満たす場合に、農作物が生育不良であると判定する請求項１に記載の施肥機制御システム。 The fertilization machine control system according to claim 1, wherein the growth defect determination unit analyzes the image and determines that the crop is growth defect when the color satisfies a predetermined condition.
- 前記位置推測手段は、ドローンのGPS、撮影高度、撮影角度、撮影向き、から前記生育不良と判定された農作物の位置を推測する請求項１に記載の施肥機制御システム。 The fertilization machine control system according to claim 1, wherein the position estimation means estimates the position of the crop determined to be the poor growth from the drone GPS, the photographing height, the photographing angle, and the photographing direction.
- 前記施肥機制御手段は、前記推測された位置に、トラクターを制御して施肥させる請求項１に記載の施肥機制御システム。 The fertilization machine control system according to claim 1, wherein the fertilization machine control means controls and fertilizes a tractor at the estimated position.
- 前記推測された位置に、ドローンを再度飛行させて再度撮像するように制御するドローン制御手段を備える請求項１に記載の施肥機制御システム。 The fertilizer applicator control system according to claim 1, further comprising drone control means for controlling the drone to fly again and image again at the estimated position.
- 施肥機を制御して、生育不良の農作物に施肥させる施肥機制御方法であって、
ドローンのカメラで撮像された画像を取得する画像取得ステップと、
前記画像を解析して、農作物の生育不良を判定する生育不良判定ステップと、
前記生育不良と判定された農作物の位置を推測する位置推測ステップと、
前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御ステップと、を備える施肥機制御方法。 It is a fertilization machine control method which controls a fertilization machine to fertilize crops with poor growth,
An image acquisition step of acquiring an image captured by the drone camera;
A growth defect determination step of analyzing the image to determine growth defects of crops.
A position inferring step of inferring a position of the crop which is determined to be in poor growth;
A fertilizer application control step of controlling the fertilizer application equipment to fertilize the crop at the position estimated as described above. - コンピュータに、
ドローンのカメラで撮像された画像を取得する画像取得ステップと、
前記画像を解析して、農作物の生育不良を判定する生育不良判定ステップと、
前記生育不良と判定された農作物の位置を推測する位置推測ステップと、
前記推測された位置にある農作物に施肥させるように施肥機を制御する施肥機制御ステップと、を実行させるためのプログラム。 On the computer
An image acquisition step of acquiring an image captured by the drone camera;
A growth defect determination step of analyzing the image to determine growth defects of crops.
A position inferring step of inferring a position of the crop which is determined to be in poor growth;
A fertilizing machine control step of controlling the fertilizing machine to fertilize the crop at the inferred position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2017/035306 WO2019064455A1 (en) | 2017-09-28 | 2017-09-28 | Fertilizer applicator control system, fertilizer applicator control method, and program |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2017/035306 WO2019064455A1 (en) | 2017-09-28 | 2017-09-28 | Fertilizer applicator control system, fertilizer applicator control method, and program |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019064455A1 true WO2019064455A1 (en) | 2019-04-04 |
Family
ID=65901877
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2017/035306 WO2019064455A1 (en) | 2017-09-28 | 2017-09-28 | Fertilizer applicator control system, fertilizer applicator control method, and program |
Country Status (1)
| Country | Link |
|---|---|
| WO (1) | WO2019064455A1 (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2015159739A (en) * | 2014-02-26 | 2015-09-07 | 井関農機株式会社 | seedling transplanting machine |
| JP2016144990A (en) * | 2015-02-07 | 2016-08-12 | ヤンマー株式会社 | Aerial spraying device |
-
2017
- 2017-09-28 WO PCT/JP2017/035306 patent/WO2019064455A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2015159739A (en) * | 2014-02-26 | 2015-09-07 | 井関農機株式会社 | seedling transplanting machine |
| JP2016144990A (en) * | 2015-02-07 | 2016-08-12 | ヤンマー株式会社 | Aerial spraying device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20240126258A1 (en) | Adaptive cyber-physical system for efficient monitoring of unstructured environments | |
| US10078890B1 (en) | Anomaly detection | |
| US20210153479A1 (en) | Monitoring livestock in an agricultural pen | |
| JP6495561B1 (en) | Land use determination system, land use determination method and program | |
| US11907341B2 (en) | Diagnostic assistance system and method therefor | |
| US11087132B2 (en) | Systems and methods for mapping emerged plants | |
| WO2018173622A1 (en) | Vegetation index calculation device, vegetation index calculation method, and computer readable recording medium | |
| JP2016168046A (en) | Plant disease diagnostic system, plant disease diagnostic method, and program | |
| Tausen et al. | Greenotyper: Image-based plant phenotyping using distributed computing and deep learning | |
| US20230306735A1 (en) | Agricultural analysis robotic systems and methods thereof | |
| Cai et al. | Land-based crop phenotyping by image analysis: Accurate estimation of canopy height distributions using stereo images | |
| Diago et al. | On‐the‐go assessment of vineyard canopy porosity, bunch and leaf exposure by image analysis | |
| US20210142484A1 (en) | Information processing apparatus, information processing method, and storage medium | |
| Gené-Mola et al. | Looking behind occlusions: A study on amodal segmentation for robust on-tree apple fruit size estimation | |
| Chopin et al. | Land-based crop phenotyping by image analysis: consistent canopy characterization from inconsistent field illumination | |
| Haberkern et al. | Maintaining a stable head direction representation in naturalistic visual environments | |
| Liu et al. | Development and experimental validation of a system for agricultural grain unloading-on-the-go | |
| JP7068747B2 (en) | Computer system, crop growth support method and program | |
| WO2018158821A1 (en) | Harvest information suggestion system, method and program | |
| Buayai et al. | End-to-end inflorescence measurement for supporting table grape trimming with augmented reality | |
| WO2019064455A1 (en) | Fertilizer applicator control system, fertilizer applicator control method, and program | |
| Christensen et al. | Sensing for weed detection | |
| US11967059B2 (en) | Methods and apparatus to identify headlands | |
| Altınbaş et al. | Detecting defected crops: Precision agriculture using haar classifiers and UAV | |
| Gan et al. | A prototype of an immature citrus fruit yield mapping system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17926602 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 17926602 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: JP |