JP2023179319A - Plant information collection device, plant growth state determination device, and plant growth state determination program - Google Patents

Abstract

To accurately determine plant growth conditions and fruit yields.SOLUTION: After sunset, target plants such as fruit trees are imaged from the ground side while illuminated from the ground side. Imaging the target plants from the ground side, rather than from an aerial view, can avoid the reflections of weeds or the like on the ground, and imaging them with illumination at night can also avoid shadowy images due to backlight. When capturing visible images of floral conditions in fruit trees and cedar etc., a set of an RGB camera and LED light is used, and when capturing multispectral images to determine the vegetative distribution and vitality, a set of an MS camera, red light, and near-infrared light is employed. One way to estimate fruit yield from visible images is, for example, counting the number of pear flower clusters from visible images of pears, acquiring the criteria that define the number of harvested fruits per cluster, and then estimating the yield.SELECTED DRAWING: Figure 1

Description

The present invention relates to a plant information collection device, a plant growth state determination device, and a plant growth state determination program, and for example, to a plant information collection device that collects information for determining the growth state of plants.

When determining the growth status of agricultural crops, fruits, etc., it has been widely used to visually observe the current situation and use empirical rules, but visual judgment is a large burden in terms of labor and cost.
Therefore, recently, technology has been developed to collect various information about the growth status of fruit trees, etc. by analyzing images taken of the condition of target fruit trees, etc.
For example, the yield of agricultural products is estimated using satellite images (Patent Document 1), or the yield is estimated using an aerial image taken of a field that is a survey target (Patent Document 2).

However, when determining the growth status of a plant using image information obtained by photographing the plant, conventional methods, including techniques described in patent documents, use images obtained by photographing the target plant from above.
For this reason, the photographed image includes, for example, weeds on the ground in addition to the target plant, so there is a problem in that it is not possible to accurately determine the state of growth.
Furthermore, when photographing a forest from above, only the upper parts of the trees could be photographed.

On the other hand, conventionally, when estimating the yield based on the growth status of fruit, a plant in a flowering state is photographed, the number of flowers is counted from the photographed image, and the yield is estimated based on the total number of flowers counted.
However, since the actual yield is the amount of fruit left after flower thinning and fruit thinning, the harvest amount was not estimated to reflect flower thinning and fruit thinning.

Japanese Patent Application Publication No. 2011-167163 JP 2015-49 Publication

The first purpose is to collect plant information that allows more accurate judgment of the growth status of plants.
The second objective is to estimate a more appropriate fruit yield based on the flowering state.

In order to achieve the above first object, the present invention provides an illumination means for illuminating a plant from the ground side, a photographing means for photographing the growing state of the plant illuminated by the illumination means from the ground side, and the photographing means A plant information collection device comprising: a position information acquisition means for acquiring photographing position information of an image photographed by the photographing means; and an output means for outputting the image photographed by the photographing means and the acquired photographing position information. I will provide a.

According to the present invention, the illumination means for illuminating plants from the ground side and the photographing means for photographing the growing state of the illuminated plants from the ground side are provided, so that the illuminated plants can be viewed from the ground side after sunset. By taking pictures, it is possible to collect plant information that can be used to more accurately judge the growth status of plants.

1 is a diagram showing the external configuration of a plant information collection device. FIG. 2 is a layout diagram showing the layout relationship between an information collection unit and a control unit. FIG. 2 is an explanatory diagram showing a shooting range of a tilted camera and tilt angles α and β. FIG. 2 is an explanatory diagram showing the configuration of a control unit and a part of its controlled objects. It is a flowchart showing a processing procedure for collecting plant information. FIG. 2 is an explanatory diagram showing the photographing point of an RGB camera and the photographing range of the photographed images a and b in Pear Garden. FIG. 2 is an explanatory diagram showing a visible image taken with an RGB camera. FIG. 2 is an explanatory diagram showing the configuration of a plant growth state determination device (harvest amount estimation device) that determines the growth state of a plant. FIG. 2 is an explanatory diagram showing determination conditions for estimating the yield of fruit from flowers shown in a visible image. FIG. 2 is an explanatory diagram schematically showing the blooming state of a pear. It is a flow chart showing growth state determination processing. It is a flowchart showing state determination processing (subroutine).

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a plant information collection device and a plant growth state determination device (harvest amount estimation device) of the present invention will be described in detail below with reference to FIGS. 1 to 12.
(1) Overview of the embodiment In the plant information collection device of the present embodiment, a target plant such as a fruit tree is photographed from the ground side after sunset (preferably at night, the same applies hereinafter) with the target plant such as a fruit tree illuminated from the ground side. It is something to do.
In this way, by photographing the target plant from the ground side rather than from above, you can avoid the reflection of weeds on the ground, and by photographing at night under lighting, you can reduce the amount of shadows caused by backlighting. becomes possible to avoid.

The plant information collection device 1 has a lighting and a camera arranged on a base 2 facing upward, and a moving part (unmanned vehicle (UGV)) 3 moves back and forth within the target field or moves in a grid pattern. (move back and forth vertically and horizontally) while taking pictures at predetermined intervals.
Lights and cameras are included, and the lights and cameras are used depending on the condition of the plants being judged.
For example, when photographing the state of flowers such as fruit trees and cedar trees as visible images, a set of an RGB camera for photographing visible images and an LED light is used.
On the other hand, when taking multispectral images to understand the distribution and activity of vegetation, a multispectral (hereinafter referred to as MS) camera, a red light (Red) light, and a near-infrared light (NIR) light are used. set is used.
In a plant growth state determination device (harvest amount estimation device), as an example of estimating the fruit yield from a visible image, for example, the number of pear flower clusters is calculated from a visible image of a pear, and the number of fruits harvested in one flower cluster is determined. Obtain the judgment conditions to estimate the yield.

(2) Details of the embodiment FIG. 1 is a diagram showing the external configuration of a plant information collection device.
As shown in FIG. 1, the plant information collection device 1 includes a base 2, a moving section 3, an information collection section 4, a control section 5, a power source 6, and other sections.
The base 2 is a part on which an information collection section 4 and a control section 5 are placed, and can be replaced with an information collection section 4 according to the information to be collected (visible image, MS image).

The moving unit 3 serves as a moving means for moving the base part on the ground side of the plant by moving in a predetermined direction within a predetermined area between the target plant (for example, a pear tree in a pear orchard) and the ground. Function.
The moving unit 3 of this embodiment is configured by an unmanned ground vehicle. The moving unit 3 performs various operations such as forward and backward movement and U-turns under the control of the control unit 5, thereby reciprocating or moving in a lattice pattern within a predetermined coordinate area. In addition, in the case of an area where target plants are planted in a grid pattern at predetermined intervals, such as a pear orchard or a vineyard, the moving unit 3 moves back and forth between each row and each column where target plants are planted. Moving.
The moving unit 3 is equipped with four tires for moving the base 2, but instead of tires, it may be equipped with caterpillars for adapting to various fields.

The moving unit 3 of this embodiment is composed of an unmanned vehicle, but it is also possible to use a drone that flies unmanned or a wagon that moves the base 2 manually. .
In either case, the information collecting section 4 is arranged so as to be able to collect (photograph) information above the base 2.
When a drone is used as the moving unit 3, flight control is executed by the control unit 5 so that the drone flies below the target plant (on the ground side).
On the other hand, when a wagon is used, three or more tires or caterpillars are arranged to move the wagon on the ground, and the wagon is moved by pushing or pulling it with human power. In this case, a handle is provided for moving the wagon.

The information collection unit 4 includes, as equipment for collecting plant information, a camera that photographs the target plant, a set of lighting devices that illuminates the photographed area, and a position detection device 40 that acquires photographing position information. .

The position detection device 40 acquires position information with higher accuracy than the position information detected by the GPS receiving device included in the RGB camera 41 and the MS camera 46, that is, at the cm level. The position detection device 40 includes a GNSS (Global Navigation Satellite System) receiving section and a reference station receiving section (not shown).
The reference station receiving unit receives from the reference station position information of the reference station that has been accurately surveyed in advance and GNSS observation data received by the reference station as reference station data.
Based on the observation instruction from the control unit 5, the position detection device 40 determines and controls the position information GN of the plant information collection device 1 from the observation data by the GNSS receiving unit of the own device and the reference station data received from the reference station. The control unit 5 stores the position information GN in the RAM.

Note that the reference station may calculate error information for the GNSS observation data from its own position information and GNSS observation data, and transmit it to the position detection device 40. The position detection device 40 corrects the observation data by the GNSS receiving unit using error information received from the reference station to obtain correct position information GN and supplies it to the control unit 5.

The camera and illumination device of the information gathering section 4 are arranged upward in order to photograph and illuminate the target plant from the ground side.
The camera and lighting device are used according to the condition of the plant to be determined (plant information to be collected).
For example, when photographing the state of flowers such as fruit trees and cedar trees as visible images (RGB images), a set of an RGB camera and an LED light is used.
On the other hand, when taking multispectral images to understand the distribution and activity of vegetation, a multispectral (hereinafter referred to as MS) camera, a red light (Red) light, and a near-infrared light (NIR) light are used. set is used.

The power source 6 is disposed at the bottom of the base 2 and supplies power as a power source for driving the moving section 3, as a power source for image capturing and illumination by the information collecting section 4, and as a power source for control by the control section 5. be done.
It should be noted that the power source 6 can be arranged separately into two, a power source for lighting and a power source other than that, or it can be further divided into three or more.

FIG. 2 shows the arrangement relationship between the information collection section 4 and the control section 5.
As shown in FIG. 2, on the base 2, when the moving direction of the moving unit 3 is set to the upper side of the drawing, an information collecting unit 4 is installed on the front side (upper side of the drawing), and a control unit 5 is installed on the rear side (lower side of the drawing). is located.
A lighting switch 50 is disposed at a predetermined position of the control unit 5, for example, on the front right side in the direction of travel. Details of the control unit 5 will be described later.

FIG. 2A shows the arrangement of the information collecting section 4 when photographing the flowering state of the target plant as a visible image.
As shown in FIG. 2(a), the information collection unit 4 includes a position detection device 40, a set of an LED light 42A that illuminates an RGB camera 41A and its photographing range, and a set of LED lights 42A that illuminates an RGB camera 41B and its photographic range. A set of LED lights 42B is arranged.
The set of RGB camera 41A and LED light 42A has an optical axis passing through the center of its imaging range and illumination range at a predetermined angle α (in this embodiment α = 30 degrees), thereby photographing and illuminating an oblique area including the vertically upward direction.
On the other hand, in the set of RGB camera 41B and LED light 42B, each optical axis is inclined at a predetermined angle α (α=30 degrees in this embodiment) to the right side (diagonally right side) in the direction of travel with respect to the vertical line. This arrangement allows imaging and illumination of the right side area including the vertically upper area.

Although the LED lights 24A and 24B provide illumination using visible light, they do not necessarily need to be LED lights, and it is also possible to emit visible light using other illumination means.
Note that the symbols A and B indicate the direction of inclination of the optical axis with respect to the vertical line, and the RGB cameras 41A and 41B and the LED lights 42A and 42B used are the same, respectively. Therefore, when explaining matters that are common regardless of the direction of inclination of the optical axis, the symbols A and B will be omitted, such as the RGB camera 41 and the LED light 42 used (hereinafter, regarding other drawings). is the same).

FIG. 2(b) shows the arrangement of the information collecting section 4 when photographing the growth status of the target plant as an MS image.
As shown in FIG. 2(b), a position detection device 40 is similarly arranged in the information collection unit 4, and red lights 47A, 47B that illuminate MS cameras 46A, 46B and their photographing ranges, Near-infrared (NIR) lights 48A and 48B are provided.
The MS camera 46A, red light light 47A, and near-infrared light light 48A are arranged such that their optical axes are inclined forward by a predetermined angle β with respect to the vertical line.
On the other hand, the MS camera 46B, the red light light 47B, and the near-infrared light light 48B are arranged such that their optical axes are inclined to the right by a predetermined angle β with respect to the vertical line.

In this way, two sets of the RGB camera 41 and the LED light 42, the MS camera 46, the red light light 47, and the near-infrared light 48 are arranged, one for diagonally forward and one for diagonally to the right. , it becomes possible to efficiently photograph the photographing range.
Note that the inclination directions can also be diagonally forward and diagonally left, diagonally backward and diagonally right, and diagonally rear and diagonally left.

The RGB camera 41 and the MS camera 46 photograph the illuminated target plant at predetermined intervals (every predetermined time t or/and every predetermined moving distance m) under the control of the control unit 5, and capture the photographed image data (visible image , MS images) are supplied to the control unit 5.
Images captured by the RGB camera 41A and MS camera 46A are saved as a images, and images captured by the RGB camera 41B and MS camera 46B are saved as b images.
In addition, the MS camera 46 outputs an MS image consisting of a set of a red light image using a red light (Red) light and a near-infrared light image using a near-infrared light (NIR) light to the control unit 5 as an a-image every time the image is captured. supplied to
Furthermore, the RGB camera 41 and the MS camera 46 have a built-in GPS device, acquire the photographing position GK at the time of photographing, and supply it to the control unit 5 together with the photographed image.

Here, the reason for tilting the optical axis of the camera and illumination and the tilt angles α and β will be explained.
In the plant information collection device 1 of this embodiment, the target plant is photographed from the ground side, rather than from the sky, where the distance to the target plant can be freely set. (less than).
Therefore, in this embodiment, by tilting the optical axis of the camera with respect to the vertical line by the inclination angles α and β, it is possible to widen the photographing range. Furthermore, it becomes possible to include more feature points (for example, branching points between branches, nodes, etc.) in one image, and it becomes possible to improve processing accuracy and speed when composing images.
The inclination angles α and β are determined according to the angle of view of the camera to be used so that the angles are located within a range of 50 mm to 400 mm inside from the end face of the angle of view.

FIG. 3 is an explanatory diagram showing the photographing range of the tilted camera and the tilt angles α and β.The RGB camera 41A and the MS camera 46A tilted diagonally forward are on the right side in the direction of movement of the plant information collection device 1. The RGB camera 41B and the MS camera 46B tilted diagonally to the right are shown as seen from the rear of the plant information collection device 1.
In FIG. 3, as an example, a pear orchard is targeted, and the photographing range is shown when the height from the camera to the pear trellis is h=1500 mm.
In Fig. 3, the vertical line is denoted by V, the optical axis of the camera is denoted by S, the range line indicating the end of the photographing range in the tilt direction of the camera is denoted by RR, and the range line on the opposite side is denoted by RL. represents.

3(a) shows the photographing range of the RGB camera 41, and FIG. 3(b) shows the photographing range of the MS camera 46.
Since the angle of view of the RGB camera 41 is as wide as 82 degrees, the optical axis S is inclined at an inclination angle α=30 degrees with respect to the vertical line V, as shown in FIG. 3(a). Due to this inclination, the distance of the photographing range in the direction of inclination can be extended to approximately 4699 mm, and an area corresponding to a distance of 11 degrees (approximately 297 mm) in terms of the angle of view is secured on the opposite side of the direction of inclination.
On the other hand, since the angle of view of the MS camera 46 is as narrow as 42 degrees, the optical axis S is inclined at an inclination angle β=20 degrees, as shown in FIG. 3(b). Due to this inclination, the distance of the photographing range in the direction of inclination can be extended to about 1571 mm, and an area corresponding to a distance corresponding to 4 degrees of angle of view (about 111 mm) is secured on the opposite side of the direction of inclination.
Note that in this embodiment, the inclination angle of the optical axis is set to α = 30 degrees and β = 20 degrees, but the inclination angles α and β depend on the angle of view of the camera used, and the shooting range includes the area above the vertical line. The inclination angles α and β are determined.

Returning to FIGS. 1 and 2, the control unit 5 controls various operations of the plant information collection device 1.
This control section 5 is provided with a lighting switch 50. The lighting switch 50 of this embodiment is connected between the lighting (LED light 42, red light light 47, and near-infrared light 48) of the information collection unit 4 disposed on the base 2 and the power source 6. The lighting is then turned on and off based on user operations.
Note that the control unit 5 may automatically control the power on and off of the illumination in conjunction with the start of photographing processing by a photographing program to be described later. In this case, the lighting switch 50 is not necessary.

FIG. 4 shows the configuration of the control unit 5 and some of its controlled objects.
As shown in FIG. 4, the control section 5 includes a CPU 51, a ROM 52, a RAM 53, an input/output section 54, a storage section 55, and a communication control section 56.
Further, the position detection device 40, RGB camera 41, and MS camera 46 that are controlled by the control unit 5 are connected to the bus line of the control unit 5, and drive control signals are supplied to the moving unit 3 (not shown).

The CPU 51 performs various information processing and control such as photographing processing in this embodiment according to programs stored in various storage units such as the ROM 52 and the storage unit 55.
The ROM 52 is a read-only memory in which various programs (system boot program, basic programs such as an OS (Operating System), etc.) and basic data for the CPU 51 to perform various controls and calculations are stored in advance.

The RAM 53 is a random access memory used by the CPU 51 as a working memory. In this RAM 53, various areas can be secured for performing various processes such as photographing processing according to this embodiment.
Specifically, in the RAM 53, photographing information 531, visible images 532, MS images 533, position information GN 534, etc. are secured as a temporary storage area.

The photographing information 531 stores various conditions necessary for photographing. That is, the photographing areas in various fields such as pear orchards specified by the user, the movement route within the photographing area, the type of photographed image (visible image or/and MS image), the photographing interval s, etc. are stored.
The shooting area and moving route of the shooting information 531 are read from the moving area 552 of the storage unit 55, but they can also be received by the communication control unit 56 or specified by the user by operating the input/output unit 54. . The imaging area is specified as an area surrounded by a plurality of position coordinates (x, y), and is necessary when the moving unit 3 is composed of the unmanned vehicle of this embodiment or a drone. In the case of a drone, the area is specified by coordinates (x, y, z) including height information. On the other hand, when the moving unit 3 is configured with a wagon that moves the base 2 manually, the photographing information 531 is not used.
The type of photographed image is determined before photographing based on the type of camera (RGB camera 41 and/or MS camera 46) arranged in the information collection unit 4, and is stored in the photographing information 531.
The photographing interval s is the interval at which the target plant is photographed, and is determined by the angle of view of the RGB camera 41 and MS camera 46 used. The photographing interval s is specified by the moving distance and moving time of the moving unit 3.

A visible image captured by the RGB camera 41 and the like are stored in the visible image 532.
The MS image 533 stores a Red image using red light, an NIR image using near-infrared light, and the like as MS images captured by the MS camera 46.
FIG. 4B shows data stored in the visible image 532 and MS image 533.
As shown in FIG. 4(b), images shot by the RGB camera 41 and the MS camera 46 (hereinafter simply referred to as both cameras 41 and 46) are managed by shooting numbers, and the shooting positions GK, a for each shooting number are The image and b image are saved.

The photographing number is assigned each time a photographing instruction is issued from the CPU 51 to each camera 41, 46.
The photographing position GK is data of a photographing position detected by a GPS device included in each camera 41, 46. In this embodiment, the photographing position GK is acquired from the RGB camera 41A and the MS camera 46A, but it is also possible to acquire it from the other camera B, and the intermediate point between the positions acquired from both cameras A and B is also possible. may be stored as the photographing position GK.

The a image is diagonally forward image data taken by the RGB camera 41A and the MS camera 46A, and the b image is diagonally right image data taken by the RGB camera 41B and MS camera 46B. Since images a and b are shot at the same time, they are managed using the same shooting number.
Note that in the a image and the b image of the MS image 533, a set of a red light image produced by a red light light and a near infrared light image produced by a near infrared light light is stored.

As the position information GN 534, accurate position information GN supplied from the position detection device 40 every time an image is taken by the RGB camera 41 or the MS camera 46 is stored.
The position information GN is stored and managed using the same imaging number as the imaging number stored in the visible image 532 and MS image 533.

The input/output unit 54 inputs and outputs data, and includes, for example, a display, a touch panel, a USB terminal, and the like.
As for the display, a liquid crystal display is used as an output device, and input results from the input supply, map information representing the photographing range by the RGB camera 41 and the MS camera 46, photographed images, etc. are displayed. The display also displays images of various operation keys (numeric keys, operation icons, etc.) for performing various necessary operations from the touch panel.
A touch panel is an input device placed on the surface of a display, and it identifies the user's touch position corresponding to the various operation keys displayed on the screen and inputs the operation key displayed corresponding to the touch position. accept.

The USB terminal is an interface for connecting a USB memory as an external storage device.
The USB memory connected to the USB terminal is used to store data to be read into the storage unit 55 or RAM 53 of the plant information collection device 1, or to store and output data of photographic information 531 to the outside.
For example, the image data and position information GN data stored in the visible image 532, MS image 533, and position information GN 534 are output and stored in a USB memory from the USB terminal as collected plant information.

The storage unit 55 includes a readable and writable storage medium, and a drive device for reading and writing various information such as programs and data to and from the storage medium. As the medium used in the storage unit 55, a semiconductor storage device such as an SSD (Solid State Drive) is used, but a hard disk, a magneto-optical disk, etc. can also be used.
The storage unit 55 stores a plant information collection program 551, a movement area 552, and collected plant information 553.
The plant information collection program 551 includes a movement control program that controls the moving unit 3 to move back and forth and/or move in a grid pattern within the range of the photographic area stored in the photographic information 531, and a movement control program that controls the moving unit 3 to move in a reciprocating manner and/or to move in a grid pattern. A photographing program and the like for photographing with the RGB camera 41 and the MS camera 46 every time and storing the photographed images are stored.
Regarding the movement control program, when an unmanned vehicle is used as the moving unit 3, the program runs on the ground, and when a drone is used, the program flies above the ground by a predetermined distance (for example, 30 cm above the ground). The program is saved. Note that if a wagon that is moved manually is used as the moving unit 3, the moving program is not necessary.

The moving area 552 stores map data for specifying a shooting area, a shooting area consisting of coordinates (x, y) of the area, and information about a moving route within the shooting area.
This information is acquired from a USB memory connected to the input/output unit 54 or an external device via the communication control unit 56 and is stored in the movement area 552.
Since the moving unit 3 of this embodiment is composed of an unmanned vehicle that runs on the ground, the imaging area is specified in an area surrounded by a plurality of position coordinates (x, y).
However, it is also possible to display a map image of the field on the display and specify the shooting range and travel route from the touch panel. In this case, the CPU 51 stores the specified photographing range and moving route in the moving area 552.
Note that when the moving unit 3 is composed of a drone, the shooting area is designated by coordinates (x, y, z) including height information from the ground over which the drone flies. Furthermore, if the moving unit 3 is configured with a wagon that moves the base 2 manually, the moving area 552 is not used.

The collected plant information 553 includes a visible image 532, an MS image 533, position information GN 534, and photographing information 531 (moving area 552) after the photographing in all the photographing areas stored in the photographing information 531 is completed. Saved as information.
Note that the plant information collecting device 1 of this embodiment does not judge the estimated fruit yield from the plant or the growing state of the plant based on the collected plant information, but stores the plant information on an external device such as a USB memory. Since the collected plant information 553 is configured to be output to the storage unit 55, it is also possible not to store the collected plant information 553 in the storage unit 55.

The communication control unit 56 is a control device that connects a portable terminal device such as a tablet or various external electronic devices such as a personal computer to a network by wireless communication.
The communication control unit 56 is capable of accessing the plant information collection device 1 from these various external electronic devices, and receives information such as photographic areas and movement routes from the external electronic devices, and collects data from the external electronic devices. This is used when transmitting plant information.

Next, processing operations for collecting plant information by the plant information collecting device 1 will be described.
FIG. 5 is a flowchart showing the processing procedure for collecting plant information.
First, the user prepares the plant information collection device 1 (step 10).
That is, the user sets the information collection unit 4 corresponding to the plant information to be collected on the base 2. That is, when photographing the flowering state of the target plant as a visible image, the information collecting unit 4 described in FIG. The information collecting unit 4 described above is set on the base 2 of the plant information collecting device 1.
Then, the plant information collecting device 1 with a predetermined information collecting section 4 set therein is prepared in the field to be photographed after sunset.
Note that the purpose of shooting after sunset is to avoid backlighting due to sunlight during shooting, so at least the shooting should be started after sunset, and the setting of the information gathering section 4, etc. should be done before sunset. is possible.

In this embodiment, an automatic vehicle is used as the moving unit 3, but if the moving unit 3 is configured with a wagon that moves the base 2 manually, the preparation of the plant information collecting device 1 (step 10) ), wrap ribbons or tapes to serve as landmarks around plants, pillars, etc. in the area (for example, the four corners) where you plan to photograph.

After the preparation of the plant information collection device 1 is completed, the user turns on the lighting switch 50 of the control unit 5 (step 11).
As a result, the illumination LED light 42, red light light 47, and near-infrared light light 48 are turned on until the lighting switch 50 is turned off after the shooting is completed, and the lighting switch 50 is turned off. Continuously illuminate the area.
In the plant information collection device 1 of this embodiment, the light is turned on by the illumination switch 50, but the light may be turned on immediately before starting photography based on the plant information collection program 551.

Next, the user starts the plant information collection program 551 (step 12).
That is, the user specifies the plant information collection icon displayed on the display of the input/output unit 54 using the touch panel, and the CPU 51 of the control unit 5 starts the plant information collection program 551.
By starting the plant information collection program 551, information collection is performed by the CPU 51 controlling each part of the plant information collection device 1 according to the program in each subsequent process.

First, the CPU 51 acquires photographing information (step 13). That is, the CPU 51 obtains the photographing area, moving route, photographed image type, and photographing interval s as photographing information.
Specifically, the CPU 51 acquires the coordinates (x, y) specifying the photographing area and the moving route from the movement area 552 of the storage unit 55 as collected information, and stores them in the photographing information 531 of the RAM 53.
Further, the CPU 51 saves a visible image (RGB image) in the photographing information 531 if the RGB camera 41 is disposed in the information collection unit 4, and an MS image as the photographed image type if the MS camera 46 is disposed. Note that if both cameras 41 and 46 are arranged, the CPU 51 stores the visible image and the MS image.

In addition, the CPU 51 obtains a photographing interval s1 for a visible image (RGB image) and a photographing interval s2 for an MS image, corresponding to the type of photographed image.
The shooting interval is determined according to the angle of view of the camera used, and in the camera of this embodiment, the shooting interval s1 of the RGB camera 41 with a wide angle of view is 2 m, and the shooting interval s2 of the narrow MS camera 46 is 1 m.
However, in this embodiment, since the moving unit 3 is moved at a predetermined moving speed (=1 m/s), the imaging interval is acquired as the imaging interval for the visible image, t1 = 2 seconds, and for the MS image, the imaging interval time t2 = 1 second. do.
Note that it is also possible to obtain s1 and s2 as the photographing interval, but by obtaining the times t1 and t2, it becomes easier to determine whether the photographing interval has been reached.

Thereafter, the CPU 51 moves the moving unit 3 to the photographing position according to the acquired movement route (step 14).
In this embodiment, as described above, since the moving unit 3 is continuously moving at a predetermined speed along the moving route, the CPU 51 determines the shooting interval time t1, t2 from the previous shooting while moving the moving unit 3. It is determined that the photographing position has been reached every time .

The CPU 51 performs photographing using the RGB camera 41 and the MS camera 46 at the photographing position of the moving unit 3 (step 15).
Specifically, if the RGB camera 41 is arranged in the information collection unit 4, the CPU 51 instructs to take a visible image every two seconds. Based on this shooting instruction, the CPU 51 acquires the a image shot by the RGB camera 41A, the b image shot by the RGB camera 41B, and the shooting position GK detected by the RGB camera 41A at the time of shooting, and for each shooting instruction. The assigned photographic number is attached and stored in the visible image 532 of the RAM 53 (see FIG. 4(b)).

Further, if the MS camera 46 is arranged, the CPU 51 instructs to take an MS image every second. Based on this shooting instruction, the CPU 51 acquires the a image shot by the MS camera 46A, the b image shot by the MS camera 46B, and the shooting position GK detected by the MS camera 46A at the time of shooting, and for each shooting instruction. It is stored in the MS image 533 of the RAM 53 with the assigned imaging number.
As described above, a set of images a and b of the MS image 533 includes a red light image illuminated by the red light light 47 and a near-infrared light image illuminated by the near-infrared light 48.

Note that when both the RGB camera 41 and the MS camera 46 are arranged in the information collection unit 4, the CPU 51 simultaneously instructs the RGB camera 41 and the MS camera 46 to take pictures every 2 seconds, and in the middle (1 second later). ), the MS camera 46 is instructed to take an individual photograph. In this case, MS images are saved with consecutive shooting numbers, and RGB images are saved with consecutive odd-numbered shooting numbers.
Although it is possible to give the shooting instruction to the RGB camera 41 and the shooting instruction to the MS camera 46 at different timings, when the movement of the moving unit 3 is temporarily stopped for shooting, it is best to give the shooting instructions at the same timing. Preferable for shortening.

In the plant information collection device 1 of this embodiment, an a image and a b image are simultaneously shot by the RGB cameras 41A and 41B in one shooting, and are saved with the same shooting number. Also, images a and b are simultaneously photographed by the MS cameras 46A and 46B and saved with the same photographing number.
FIG. 6 shows, as an example, the photographing point of the RGB camera 41 and the photographing range of the photographed images a and b in Pear Garden.
As shown in FIG. 6, in the pear orchard, a plurality of pear trees PL are regularly planted in a grid pattern, and the plant information collection device 1 reciprocates between each row of pears along a path indicated by a long arrow. Photographing is performed by the RGB camera 41 at photographing positions p1, p2, . . . on this moving route.
Note that for the pear tree PL, only the position of the tree is represented by a circle in order to explain the photographing ranges a1, b1, etc.

Further, as shown in FIG. 6, at the photographing position p1, the visible image (a image) photographed by the RGB camera 41A is captured in the photographing range a1 indicated by the solid line, and the visible image (b image) photographed by the RGB camera 41B is similarly photographed by the solid line. It is represented by range b1.
On the other hand, at the next shooting position p2 moved 2 m (2 seconds) from the shooting position p1, an image a taken by the RGB camera 41A is represented by a shooting range a2 indicated by a dotted line, and an image b taken by the RGB camera 41B is represented by a shooting range b2 similarly indicated by a dotted line. There is.
In this way, by tilting the two RGB cameras 41A and 41B diagonally forward and diagonally to the right, respectively, it becomes possible to photograph wide areas (a1+b1) and (a2+b2) in one photographing.

FIG. 7 shows a visible image taken by the RGB camera 41.
FIG. 7A is a visible image (image a) taken by the RGB camera 41A, and since the image is taken diagonally forward, more of the branch tips than the trunk of the pear tree are photographed.
On the other hand, FIG. 7(b) is a visible image (b image) taken by the RGB camera 41B, and since the image is taken diagonally to the right, the area around the trunk of the pear tree on the right side of the plant information collection device 1 is taken. .

Returning to the flowchart of FIG. 5, the CPU 51 acquires the position information GN from the position detection device 40 (step 16).
That is, the CPU 51 instructs the position detection device 40 to perform observation at the timing when the RGB camera 41 and MS camera 46 are instructed to take a photograph.
Based on this observation instruction, the position detection device 40 observes and outputs accurate position information GN of the plant information collection device 1, and the CPU 51 acquires the output position information GN and stores it in the position information GN 534 of the RAM 53. .
The position information GN is managed and stored using the same shooting number as the shooting number for the shooting instruction issued at the same time as the shooting instruction.

When the photographing and data storage are completed, the CPU 51 determines whether all photographing has been completed within the photographing area stored in the photographing information 531, based on whether the end point of the moving route has been reached (step 17).
If the photographing is not completed (step 17; N), the CPU 51 returns to step 14 and continues moving the moving unit 3 and photographing.

On the other hand, if photographing has been completed (step 17; Y), the CPU 51 outputs collected plant information based on photographic data and the like (step 18).
The collected plant information output here is all the image data and position information GN data stored in the photographing information 531 (or moving area 552), visible image 532, MS image 533, and position information GN 534 of the RAM 53.
The collected plant information can be output to a USB memory connected to the input/output section 54, an external device connected via the communication control section 56, or the like. The collected plant information in the RAM 53 is also stored in the collected plant information 553 in the storage unit 55.
After outputting the collected plant information, the CPU 51 ends the plant information collection process.

Next, a description will be given of how to determine the growth status of a target plant using collected plant information (visible images, MS images, etc.) output from the plant information collection device 1.
FIG. 8 shows the configuration of a plant growth state determination device (harvest amount estimation device) that determines the growth state of plants.
As shown in FIG. 8, the plant growth state determination device 100 includes a CPU 110, a ROM 120, a RAM 130, a USB terminal 140, an input section 150, a display device 160, a communication control section 170, and a storage section 180.

The CPU 110 performs various information processing and control, such as processing for determining plant growth information in this embodiment, according to programs stored in various storage units such as the ROM 120 and the storage unit 180.
The ROM 120 is a read-only memory in which various programs (system boot program, basic programs such as an OS (Operating System), etc.) and basic data are stored in advance for the CPU 110 to perform various controls and calculations.

RAM 130 is a random access memory used by CPU 110 as working memory. This RAM 130 can secure various areas for performing various processes such as photographing processing according to this embodiment.
Specifically, in the RAM 130, visible images 132, MS images 133, position information GN 134, feature points 136, composite images 137, determination results 138, etc. are secured as temporary storage areas.
The visible image 132, the MS image 133, and the position information GN 134 are collected plant information collected to determine the growing state of the plant. This collected plant information is information corresponding to the photographing information 531 (or moving area 552), visible image 532, and MS image 533 of the plant information collecting device 1, and is stored in the storage unit 180 in the process of determining the growing state of the collected plants. The collected plant information 182 is read out and saved. One or both of the visible image 132 and the MS image 133 are stored in the RAM 130.

The position information GN 134 is used when combining a plurality of visible images 132 and MS images 133 corresponding to the photographing position.
The positional information GN 134 is information representing a more accurate photographing position than the photographing position GK included in the visible image 132 or the MS image 133, and is the positional information GN detected by the position detecting device 40 of the plant information collecting device 1.
Since the position information GN 134 is managed with the same shooting number as the shooting number of each shot image stored in the visible image 132 and MS image 133, it is overwritten with the position information GN of both images 132 and 133 with the same shooting number. be done.

The feature points 136 are stored as feature points extracted from each photographed image (visible image, MS image) taken at a plurality of locations in one field.
The feature points 136 are feature points such as branch points F and nodes H extracted from each captured image by image processing.
The feature points 136 are used when combining each captured image.

The composite image 137 is a composite image of the visible image 132 and the MS image 133, which are composited based on the feature points 136 extracted from each captured image.
The composite image 137 of this embodiment is a composite image of the entire field where the collected plant information was collected, but depending on the field, the composite image 137 may be created for each pre-divided area. It is possible.

As the determination result 138, the result of determining the growth information of the plant depicted in the image using the composite image 137 is stored.
For example, the harvest amount of fruits determined from the state of flowers in visible images, the normalized vegetation index (NDVI) for understanding the growth status of plants from MS images, etc. are stored.

The USB terminal 140 is an interface for connecting a USB memory as an external storage device, and a USB memory in which collected plant information collected and output by the plant information collecting device 1 is stored is connected.
The input unit 150 is various input devices such as a touch panel and a keyboard arranged on the display surface of the display device 160. The input unit 150 specifies target data for determining the growth state, inputs and specifies determination information such as conditions necessary for determining the plant growth state, instructs processing to start the determination process, and performs other functions. Used for various input operations.
The display device 160 is an image output device such as a liquid crystal display. The display device 160 displays visible images before and after synthesis, MS images, NDVI images representing the distribution state of the normalized vegetation index, and the like. The display device 160 also displays various operation keys for enabling touch panel operations.

The communication control unit 170 is a control device that connects various external electronic devices such as mobile terminal devices such as tablets and personal computers to a network through wireless communication.
The communication control unit 170 of this embodiment is also used when acquiring collected plant information output from the plant information collection device 1 by wireless communication.

The storage unit 180 includes a readable and writable storage medium, and a drive device for reading and writing various information such as programs and data to and from the storage medium.
The storage medium used in this storage unit 180 is mainly a readable/writable storage medium among various storage media such as a hard disk, a semiconductor storage device such as an SSD (Solid State Drive), and a magneto-optical storage medium. You can also do this.
The storage unit 180 stores a growth state determination program 181, collected plant information 182, determination conditions 183, plant growth state determination information 184, and other programs and data.

The growth state determination program 181 is a program for determining the amount of fruit harvested, growth information of various plants such as normalized vegetation index, and the amount of pollen of cedar and cypress based on collected plant information. The growth state determination program 181 uses captured images (visible images, MS images) stored in the collected plant information 182 according to the determination target plant designated by the user and conditions selected from the determination conditions 183 described below. to determine the growth status of the target plant.
The collected plant information 182 stores collected plant information output from the plant information collection device 1. This collected plant information 182 is stored in the work area of the RAM 130 by executing the growth state determination program 181.
The plant growth state determination information 184 stores determination results such as the amount of fruit harvested. As for this determination result, the determination result 138 stored in the RAM 130 in the growth state determination process is read out and stored in the plant growth state determination information 184.

In the determination condition 183, various determination conditions used in the growth state determination program 181 are stored. Specifically, the judgment conditions 183 include judgment conditions for each fruit to estimate the yield of fruit from visible images, a calculation formula for calculating the normalized vegetation index (NDVI) from MS images, and estimation of pollen amount of cedar and cypress. Estimated conditions, etc. when doing so are saved.
The user specifies which judgment condition to use or not, but if there is no specification by the user, a predetermined default judgment condition is used.
Default conditions are predetermined as the various judgment conditions stored in the judgment conditions 183, but they can be changed or added as appropriate by the user.

Here, the judgment conditions stored in the judgment conditions 183 used in this embodiment, mainly the judgment conditions for fruits, will be explained.
In general, for various fruits such as pears, apples, peaches, and kiwis, the yield of fruits is estimated by counting the number of flowers.
However, in actual orchards, instead of letting all the flowers bear fruit, they perform flower thinning and fruit thinning, which removes flowers and small fruits midway through the harvest, to harvest the optimal fruit (large, delicious, etc.). There is.
Therefore, in this embodiment, rather than counting the number of flowers, by counting the state of flowers corresponding to flowers and fruits left for harvesting, a more accurate harvest amount is estimated as a result of determining the plant growth state. .
In the case of fruit trees, the determination conditions 183 define conditions for estimating the yield from the state of flowers.

FIG. 9 shows, as an example, determination conditions for estimating the yield of fruits such as pears and kiwis from the flowers of the fruit depicted in the visible image 132. (a) is the determination condition for pears, (b) represents the criteria for determining kiwifruit.
As shown in FIG. 9(a), the conditions for determining pear include "the number of fruitless flowers at the tip of a branch", "the number of fruits in the flower cluster", etc.
Here, the first condition "number of fruitless flower clusters at branch ends" is the number of flower clusters that are counted from the tip of the branch (branch tip) and is not counted as the number of harvested pears, and is the number of flower clusters that do not bear fruit.
The next condition "Number of fruits in a flower cluster" is the number of fruits scheduled to be harvested in one flower cluster, and is the number of fruits (estimated number of fruits to be harvested) counted for one flower cluster.
The "number of fruitless flowers at the tip of a branch" is defined as 2 in condition 1, 3 in condition 2, 3 in condition 3, etc., and the "number of fruits in the flower cluster" is 1 fruit in condition 1, 1 fruit in condition 2, etc. Condition 3 stipulates that there are 2 results.

Here, the relationship between the harvest of pears and the default determination conditions will be explained.
FIG. 10 schematically represents the flowering state of a pear.
As shown in FIG. 10, in a pear, a plurality of flowers bloom from one flower bud, and the flower group is called a flower cluster (KS1, KS2, etc.).
In the case of pears, for the purpose of promoting the growth of new shoots and distributing nutrients throughout the tree, in order to prevent fruit from growing on the tip of the branch (branch tip), the branch point is For n flower clusters in the direction, all flowers may be thinned.
The number of flower clusters from which all flowers are removed from the branch tips is defined as the "number of branch tip fruitless clusters."
For example, in Judgment Condition 1, the "number of branch tip fruitless flowers" is "2" (Figure 9(a)), so two flower clusters, KS1 and KS2 (Figure 10) at the tip of the branches, are subject to complete flower thinning and harvested. The estimated number is 0 (not counted).

On the other hand, if one flower cluster bears many fruits, it will not be possible to harvest large fruits. Therefore, flowers and fruits must be removed from each flower cluster at appropriate times until harvest, and the fruits that can be harvested from a single flower cluster must be removed at appropriate times until harvest. The number is defined as the "number of fruits in the flower cluster."
For example, in judgment condition 1, the "number of fruits in a flower cluster" is "1 fruit" (Figure 9(a)), so the estimated number of harvested pears for the flower cluster KS3, excluding flower clusters KS1 and KS2, which are not counted. 1 is counted.
Note that the flower cluster at the tip of a branch defined by the "number of fruitless flowers at the end of a branch" is not counted by the "number of fruits in a flower cluster".

The judgment conditions for pears specified in Figure 9(a) are just an example, and other conditions may be used, such as conditions corresponding to the target pear type (Kosui, Hosui, Nikkori, etc.) or conditions based on the pear orchard's own harvesting policy. It is now possible to configure and add items such as.
As a pear orchard's own harvest policy, for example, as a condition for harvesting extra-large pears in a specific area, even in flower clusters other than the tips, one fruit per multiple (for example, 2, 3, ... 5) flower clusters is stipulated. Or, the distance between the flower clusters to be harvested may be defined as Bcm (for example, 20 cm) or more.
In this way, by specifying the judgment conditions for each type of pear, including those shown in Figure 9(a), it is possible to more accurately estimate the yield by selecting the judgment conditions suitable for the pear orchard operated by the user. It becomes possible.

Next, the criteria for determining kiwifruit shown in FIG. 9(b) will be explained.
As shown in FIG. 9(b), "target", "number of fruit on stalk", "leaf number condition", "branch length condition", etc. are defined as criteria for kiwifruit.
Kiwi is a male and female tree, so all female flowers are required to be counted.
Kiwi flowers are recognized by image recognition, which identifies the outer shape of the flower based on its shape and white color, and further identifies female flowers based on the presence or absence of white thread-like pistils extending outward from the center. A male flower has a spherical filament with yellow pollen attached to its center.

Since kiwifruit usually have three flowers blooming on one fruit stalk, the number of kiwis left (harvested) by flower thinning and fruit thinning for one fruit stalk is defined as the "number of fruits on a fruit stalk."
Further, in the "leaf number condition", the number of fruits to be harvested relative to the number of kiwi leaves is defined. For example, in condition 1, one kiwi fruit is counted for every five leaves.
The branch length condition is the number of fruits to be harvested (counted) according to the length of the branched branches. For example, in condition 3, it is defined as 1 fruit if small, 2 fruits if medium, and 3 fruits if long. There is. Note that the distinction between small, medium, and long is also specified separately. For example, small is defined as less than 30 cm, medium is defined as 30 cm or more and less than 80 cm, and length is defined as 80 cm or more.

Although not shown, the judgment conditions 183 include various judgment conditions such as the number of fruits to be harvested relative to the number of flowers (number of flowers, number of flower clusters, fruit stalks, etc.) and counting conditions for leaves and branches. Regulations are specified for each fruit tree, such as apples and grapes.

Next, a growth state determination process (harvest amount estimation process) that uses the collected plant information collected and output by the plant information collection device 1 to determine the growth state of the plant will be described.
FIG. 11 is a flowchart showing the growth state determination process (harvest amount estimation process).
As a premise of this processing, collected plant information including visible images and MS images collected by the plant information collecting device 1 is stored in the collected plant information 182 of the storage unit 180 or a USB memory connected to the USB terminal 140, or communication control is performed. It is assumed that collected plant information can be acquired from the outside via the unit 170.
It is also assumed that the growth state determination program 181 has been activated by the user's operation on the input unit 150.

As shown in FIG. 11, the CPU 110 of the plant growth state determination device 100 acquires determination conditions (step 20). That is, the CPU 110 displays a screen for specifying a determination condition on the display device 160, and acquires the determination condition specified by the user from the determination condition 183 based on this display. Here, assuming that the user specifies the pear determination condition 1 shown in FIG. .

Next, the CPU 110 acquires photographic data (step 21).
That is, the CPU 110 acquires the collected plant information of the corresponding plant from the collected plant information 182 (or USB memory, communication control unit 170) based on the determination condition 1 specified by the user, and reads the visible image 132 and/or the visible image 132 of the RAM 130. The MS image 133 and position information are saved in the GN 134.

Next, the CPU 110 corrects the photographing positions of the visible image 132 and the MS image 133 (step 22).
That is, the CPU 110 converts the photographing position GK (see FIG. 4(b)) included in each photographed image of the visible image 132 and the MS image 133 into position information managed using the same photographing number as that of the photographed image. Change (correct) the position information GN of GN134.
Hereinafter, regarding the photographing position GK, the notation after correcting the photographing position using the position information GN will be referred to as the photographing position GN.

In this embodiment, since the accuracy of the photographing position GK acquired by the GPS device included in the camera of the plant information collecting device 1 is low and image synthesis (step 24) described below cannot be performed, the position information GN with high accuracy is corrected. ing.
However, when the photographing position GK is obtained with high precision to the extent that images can be synthesized, the position detecting device 40 is not installed in the plant information collection device 1, and the photographing position GK is used as is.
In this case, step 22 of correcting the photographing position is omitted.
Step 22 is also omitted when the collected plant information 182 with the photographing position GK corrected to the photographing position GN is acquired from the plant information collection device 1.

Next, the CPU 110 extracts feature points (step 23).
That is, the CPU 110 extracts feature points from each photographed image of the visible image 132 and the MS image 133, attaches a photographing number to the extracted feature points, and stores them as the feature points 136.
Feature points of the photographed image include, for example, nodes H of the target plant and branch points F of branches shown in FIG. 10, which are extracted by image recognition.
In addition, as the feature points to be extracted, predetermined landmarks (for example, red tape or strings) are placed on the branches and trunks of plants before photographing by the plant information collection device 1, and these landmarks are extracted as feature points. You may also do so. In this case, it is preferable that the marks are arranged at the same interval as the photographing interval by the plant information collection device 1. In the case of this embodiment, the marks are arranged at an interval of s1 = 2 m for the RGB camera 41 and 1 m for the narrow MS camera 46.

The extracted feature points are expressed by coordinates with a predetermined point in the photographed field of the photographed field as the origin, based on the photographing position GN of the target photographed image and the position of the feature point in the photographed image.
The photographing area of the field is acquired from the photographing information 531 (or moving area 552, see FIG. 4) included in the collected plant information 182.
As the origin, for example, the first photographed point (location information GN of photographing number 001) is used.

Although it is possible to extract feature points from the entire image of each photographed image, in this embodiment, the feature points are extracted from the peripheral area of the image except for the central part.
As illustrated in FIG. 6, in the photographing by the plant information collecting device 1, the photographed images are taken so that a part of the photographed images on the front, back, left and right sides overlap in the back and forth photographing and the back and forth photographing. Extract feature points as follows.
Note that the visible image 132 and MS image 133 used in this synthesis process are obtained by photographing the target plant from the ground side with the plant information collecting device 1 illuminated after sunset. Therefore, since the image is not taken in a backlit state, extraction of feature points (and image synthesis) can be easily performed.

Next, the CPU 110 performs image composition (step 24).
That is, the CPU 110 synthesizes the visible image 132 and the MS image 133 using the extracted feature points 136 of each captured image and the captured position GK.
More specifically, the CPU 110 first determines adjacent photographed images of the visible image 132 and the MS image 133 from the photographing number and the travel route at the time of photographing. Then, based on the positional relationship between the two adjacent captured images, positioning is performed so that among the feature points 136 corresponding to the imaging numbers of both of the captured images, the feature points 136 in the surrounding area where the images overlap coincide with each other.
The CPU 110 stores, in a composite image 137, a photographed image of the entire field in which all photographed images are aligned with adjacent photographed images.

Regarding the synthesis of captured images, in addition to simply compositing by aligning the captured images, it is also possible to create an ortho image that represents the entire field three-dimensionally by composing the visible image 132 and MS image 133. You may also do so.
To create an orthogonal image, SfM (Structure from Motion) software is used to create a three-dimensional orthogonal image by combining images taken at different photographing positions.

Next, the CPU 110 determines the growth state (step 25), and ends the process.
That is, the CPU 110 uses a composite image 137 of the visible image 132 and the MS image 133 to perform state determination processing regarding the growth state.
FIG. 12 is a flowchart showing the state determination process (subroutine).
The state determination process is performed according to the determination conditions acquired in step 20, but here, a case will be described in which the pear yield is estimated (determined) assuming that determination condition 1 in FIG. 9(a) is specified. .

First, the CPU 110 extracts a flower cluster (step 241).
In the case of pears, multiple flowers bloom in one flower cluster KS1, KS2, etc. (see Figure 10), and one or several pears per flower cluster are to be harvested, so the number of fruits harvested in each flower cluster is used as a judgment condition. (See FIG. 9(a)).
Therefore, the CPU 110 does not extract individual flowers from the composite image 137 obtained from the visible image 132, but extracts a flower cluster composed of a plurality of flower groups. To extract the flower cluster, various well-known image recognition techniques such as extracting the outline of the entire flower cluster can be used.

In the plant information collection device 1 of this embodiment, since the photograph is taken from the ground side with lighting after sunset, the target plant is photographed against a black background, as shown in FIG. . Additionally, weeds and the like are not captured in the background, unlike when photographing from above the target plant during the day.
In the photographed image (composite image 137) of this embodiment, there are no weeds in the background of the target plant, and the background is black with clear outlines of flowers and flower clusters, making it easy to extract flowers and flower clusters. It can be carried out.

Next, the CPU 110 counts the number of target flower clusters (step 242).
That is, the CPU 110 counts the number of target flower clusters based on the determination conditions acquired in step 20 (FIG. 11) for the flower clusters extracted in step 241.
When condition 1 in FIG. 9(a) is selected as the determination condition, the flower cluster at the tip of the branch and the flower cluster one place before it among the extracted flower clusters (in the example of FIG. 10, the flower cluster KS1 and the flower cluster KS2) The number of fruits excluded is counted as the number of target fruits.

Next, the CPU 110 estimates the harvest amount (step 243).
That is, the CPU 110 estimates the harvest amount T of the target fruit from the following equation (1), where W is the counted number of target flower clusters and U is the determination condition "number of fruits in the flower cluster".
Harvest amount T = Number of target flower clusters W x Number of flower clusters U... (1)
For example, when condition 1 in FIG. 9A is specified, the number of fruits in the flower cluster U = 1 fruit, so the CPU 110 estimates the harvest amount T = the number of target flower clusters W x 1 = W.

Finally, the CPU 110 stores the estimated harvest amount T (step 244).
That is, the CPU 110 saves the estimated pear yield T as the plant growth state determination information 184, returns to the main routine of FIG. 11, and ends the process.

The growth status determination process for estimating the yield of pears has been described above, but other fruits such as kiwis, apples, peaches, grapes, cherries, citrus fruits such as mandarin oranges, and hassaku can also be evaluated using the determination conditions specified for each fruit. The harvest amount of fruit is estimated according to the determination condition selected by the user or the determination condition determined as a default.

In addition, although a case has been described in which the visible image 532 (visible image 132) is used among the collected plant information 553 collected by the plant information collecting device 1 to estimate the yield of fruit as the growing state of the plant, It is also possible to determine the growth state of a plant using a visible image (RGB image). In this case as well, the visible image 532 (visible image 132) taken by the plant information collection device 1 after sunset of the plant from the ground side in an illuminated state is used to determine the growing state.

Next, a second embodiment using the plant information collection device 1 and the plant growth state determination device 100 will be described.
In this second embodiment, as an example other than fruits, plant information on male flowers of cedar and cypress is collected, the epiphytic status (plant growth status) of the male cedar flowers is determined, and the amount of pollen in the spring of the following year is estimated. .

Allergies caused by cedar and cypress pollen have become a major problem, and the amount of pollen dispersed is predicted to increase. Regarding this forecast, pollen counts are investigated during the period when pollen is actually scattered and used for pollen forecasting. Pollen counts are measured by humans using a microscope to count the number of pollen particles that stick to prepared specimens left outdoors, or by automatically measuring the number of pollen particles in the air by shining a laser beam onto the air sucked in by a pump. .
However, this method examines the pollen scattering situation during the pollen scattering season, and it is not possible to estimate and predict the pollen scattering situation before the pollen scattering season.

Therefore, a technology has been proposed that predicts the amount of pollen scattered in the spring of the following year based on the previous year when the pollen is scattered (Nagano Prefecture Forest Research Center Report No. 21, pp. 19-28, "Pollen Production Prediction System Dissemination Project" / 2006) December, Kondo, Michiharu).
However, with this prediction system, it was necessary to visually observe the crowns of cedar trees using binoculars to evaluate the status of male flower formation on each cedar tree, which was a time-consuming and difficult task.
Additionally, since visual observations were to be made using binoculars during the day, it was necessary to consider the positional relationship with the sun.

Therefore, in the second embodiment, as an example other than fruits, visible images of male flowers of cedar and cypress are photographed, outputted as collected plant information, and this collected information (visible images) is image-processed to increase the epiphytic growth of male flowers of cedar. It assesses the situation and estimates the amount of pollen for the following spring.
The estimation of the amount of cedar pollen using the plant information collection device 1 and the plant growth state determination device 100 according to the second embodiment will be described below, focusing on the differences from the processing described in the first embodiment.

In the second embodiment, a photograph of a cedar forest is taken using the plant information collecting device 1 equipped with the set of the RGB camera 41 and the LED light 42 described in the first embodiment.
This photographing is carried out at predetermined intervals after sunset from November to December when the male cedar flowers are established, under illumination with the LED light 42, in accordance with the photographing area and movement route specified in the movement area 552. An RGB camera 41 photographs a visible image of the cedar from the ground side.
Although it is possible to use a wagon that is moved manually or a drone that flies unmanned as the moving unit 3 of the second embodiment, an unmanned vehicle is used as in the first embodiment. However, since the vehicle moves within a cedar forest, it is preferable to use an unmanned vehicle equipped with caterpillars instead of the four tires shown in FIG.

Further, in the movement area 552 of the plant information collection device 1, a photographing area and a movement route are stored as in the first embodiment.
However, unlike the pear orchard described in the first embodiment where the target plants are regularly planted in a grid pattern, the target cedars are located at arbitrary positions. Therefore, in the second embodiment, a moving route that allows the user to photograph the cedars in the entire photographing area while avoiding the cedar trees is stored in the moving area 552.

Next, a process of acquiring the collected plant information 553 of cedar outputted from the plant information collecting device 1 and estimating the scattering state of cedar pollen using the plant growth state determining device 100 will be described.
In the second embodiment, as the determination conditions 183 stored in the storage unit 180, "number", "classification condition", "weight", "calculation formula", etc. are defined.

The "number" is the number of cedar trees to be surveyed that exist within the photographing area, and is determined to be 40. These 40 lines are predetermined 40 lines that exist within the imaging area, but it is also possible to use 40 arbitrarily extracted lines as the target. When 40 trees are determined in advance to be surveyed, an identification tag is attached to each cedar tree for image recognition to identify the cedars as survey targets.

The "classification conditions" are conditions for classifying cedar trees into A to D based on the status of male flower formation on each cedar tree. Classifications A to D are defined as follows.
Classification A: Individuals with male flowers growing on the entire crown of the cedar tree and a very high density of male flower clusters Classification B: Individuals with male flowers growing on almost the entire crown of the tree Category C: Male flowers are scattered on the crown of the tree Individuals that are epiphytic or epiphytic on a limited part of the tree crown Class D: Individuals where male flowers are not observed In this embodiment, male flower density Q determined from the crown part of the cedar tree by image recognition for classifications A to D. Determine from. The male flower density Q is the ratio of the yellow area Y, which is the male flower part, to the green area G, which is the needle leaf part, and is Q=(Y/G)×100.
Male flower density Q is classified as A if it is 40% or more, B if it is 10% or more and less than 40%, C if it is 1% or more and less than 10%, and D if it is less than 1%. It is stipulated to do so. The density range of male flower density Q corresponding to each classification is an example adopted in this embodiment, and the density range may vary depending on the conditions of each observation region (cold region, warm region, highland, flatland, etc.). It is determined for each condition 1, condition 2, etc., and is determined at step 20 (obtaining judgment conditions) in FIG.

"Weight" is a weight value given to each classification condition, and is 100 for classification A, 50 for classification B, 10 for classification C, and 0 for classification D.

The "calculation formula" is a formula for calculating the pollen production amount KH using the male flower index D.
The male flower index D represents the state of male flower formation and is calculated from the following equation (2).
Male flower index D = (Number of A x 100) + (Number of B x 50) + (Number of C x 10) + (Number of D x 0)...(2)

The pollen production amount KH is a formula for estimating the number of male flowers generated (produced) per unit area (1 square meter) in a cedar forest.For example, the following formulas (3) and (4) can be calculated by the user. It is specified in the selected judgment condition.
The first equation (3) is as follows.
Pollen production amount KH=0.1452×D+134.79…(3)
In formula (3), the coefficient of determination R{2} (R{2} is the square of R) is R{2}=0.9177, and there is a strong correlation between the male flower index D and the pollen production KH. A relationship is recognized.

The following equation (4) is as follows.
Pollen production amount KH=0.9934×E+0.5842…(4)
The coefficient of determination R{2} in equation (4) is 0.9246, and a strong correlation is recognized between male flower index D and pollen production KH in equation (4) as well.
In addition, in this formula (4), E is the second male flower index, and the value obtained by dividing the number of classification A by the number of survey subjects Z (= 40) (number of A / Z) is the A rank rate, and the following It is expressed by formula (5).
Second male flower index E = male flower index D × (1 + A rank rate)... (5)

In the second embodiment, steps 20 to 24 of the growth state determination process shown in FIG. 11 are processed in the same manner as in the embodiment.
On the other hand, in the state determination process in FIG. 12 (step 25 in FIG. 11), the CPU 110 first extracts 40 target cedar trees from the synthesized visible image.
Next, the CPU 110 determines the male flower density Q of each extracted target cedar tree by image analysis, classifies it into A to D, and counts the number of flowers in each classification A to D.
Further, the CPU 110 estimates the pollen production amount KH per unit area using equation (3) or equation (4) selected as the determination condition, stores it in the plant growth state determination information 184, and ends the process.

The estimation of the pollen production amount KH according to the second embodiment is preferably performed at a plurality of locations (a plurality of imaging regions) within a specific area, for example, on a city, town or village basis or on a prefecture basis. Then, by integrating the pollen production amounts KH calculated for a plurality of photographic regions, the pollen scattering amount for each region and region in the next spring is predicted.

In the second embodiment, a case has been described in which the pollen production amount KH of cedar is estimated, but the pollen production amount of cypress can also be determined in the same manner as for cedar.
However, as the determination condition 183, the condition obtained from the investigation results for Hinoki is used.

The case where the visible image 132 is used to estimate the harvest amount and pollen production amount of pears has been described above.
Next, as a third embodiment, determination of the growth state of a plant using the MS image 133 captured by the plant information collection device 1 will be described.
In the third embodiment, in order to enable determination of the plant growth state, the MS image 133 is used to visualize the plant growth state using an NDVI image.
That is, chlorophyll contained in plant leaves has the characteristic of absorbing red light but strongly reflecting wavelengths in the near-infrared region. Taking advantage of the characteristics of this vegetation, the normalized vegetation index (NDVI) is calculated using the values of red wavelength (Red) and near-infrared wavelength (NIR) in MS images (red light image and near-infrared light image). By creating an NDVI image that represents the distribution state of the normalized vegetation index, the level of activity of plants can be visualized by different colors and shades of NDVI values.
For example, in the NDVI image of this embodiment, the higher the normalized vegetation index, the higher the activity level of the plants, and green parts represent areas where the NDVI value is high, and red areas represent areas where the NDVI value is low.

In the third embodiment, a normalized vegetation index is calculated using the collected plant information, and by extracting the difference between the past and present from the normalized vegetation index collected on different dates and times, The location, range, and extent of changes in vegetation can be easily confirmed.
In addition, by calculating multiple times at predetermined intervals (for example, every week, every 10 days, every month, etc.), it is possible to understand changes in the growth status of plants, such as the timing and range of fertilization, harvest This is useful when understanding the timing of
According to this embodiment, since the MS image 133 collected by the plant information collection device 1 and photographed after sunset with lighting from the ground side is used, there is no soil in the background of the image, so the background It becomes possible to avoid the influence of soil.

In addition, since the photographing of the MS image 133 by the plant information collection device 1 is explained in the first embodiment, in this third embodiment, the plant growth state determination using the MS image 133 by the growth state determination program 181 will be explained. explain.
In the third embodiment, the MS camera 46 of the plant information collection device 1 produces two types of images: a red light image by red light irradiated by the red light light 47 and a near-infrared light image by irradiation by the near-infrared light 48. Photographs were taken at each shooting position.
Then, the CPU 110 performs processing from correcting the photographing position to combining the images for each of the red light image and the near-infrared light image (steps 22 to 24), and uses both the combined images in the state determination process (step 25). to create a normalized vegetation index (NDVI) image.

In the state determination process (step 25), the CPU 110 acquires the pixel value Red of the red band and the pixel value NIR of the near-infrared band for each corresponding pixel of the red light image and the near-infrared light image in the MS image 133. , the normalized vegetation index (NDVI) of each pixel is obtained from the following equation (6) and stored in the determination result 138 of the RAM 130.
NDVI=(NIR-Red)/(NIR+Red)...(6)
In equation (6), the value of NDVI at each pixel is in the range of -1 to +1.

Note that it is also possible to convert the data output range into an integer (8 bits) in the range of 0 to 200 using the following equation (7).
Integerized NDVI=(NDVI+1.0)×100...(7)

The larger the numerical value of the NDVI and integerized NDVI calculated using equations (6) and (7), the more vegetation there is.
In NDVI images (including normalized NDVI images, the same applies hereinafter), the higher the normalized vegetation index, the higher the activity of plants. Visualize.
For example, the normalized vegetation index is displayed as an image in the range of green to red, with green areas representing areas where the activated vegetation index value is high and red areas representing areas where the value is low.

Note that, in addition to the above equations (6) and (7), other definitions can also be used as indicators of vegetation.
For example, it is also possible to obtain a specific vegetation index expressed by the following equation (8) to understand and judge the growing state of plants.
Specific vegetation index = NIR/Red…(8)

Although the embodiments of the plant information collecting device 1 and the plant growth state determining device 100 of the present invention have been described above, the present invention is not limited to the described embodiments, and various types can be used within the scope of each claim. Variations can be made.
For example, in the plant information collecting device 1 of this embodiment, as shown in FIGS. 2(a) and 2(b), the information collecting unit on the base 2 The case of changing 4 (camera and lighting) has been explained.
On the other hand, a position detection device 40, RGB cameras 41A, 41b, LED lights 42A, 42B, MS cameras 46A, 46B, red light lights 47A, 47B, and near-infrared light lights 48A, 48B are arranged on the base 2. You may also do so.
This makes it possible to collect various types of information without replacing the information collecting section 4.

Furthermore, in the described embodiment, the visible image 532 and/or MS image 533 and position information GN 534 stored in the RAM 53 are outputted to an external device such as a USB memory as plant information, and included in each image data in the plant growth state determination device 100. A case has been described in which the photographing position GK (see FIG. 4(b)) is changed to position information GN indicating the accurate photographing position.
On the other hand, the CPU 51 of the plant information collection device 1 may change the photographing position GK included in the visible image 532 and the MS image 533 to the photographing position GN using the position information GN.
In this case, each time the CPU 51 acquires the position information GN from the position detection device 40 in step 16 (see FIG. 5), the CPU 51 changes the photographing position GK described in FIG. 4(b) to the photographing position GN.
Therefore, the position information GN534 is not stored in the RAM 53, and the collected plant information 553 that is output to the outside and stored in the storage unit 55 does not include the position information GN534.

Since the plant growth state determination device 100 has already acquired the collected plant information 182 (visible image 132, MS image 133) that has been changed to the photographing position GN, the process of correcting the photographing position in step 22 is not necessary. .

In addition, in the plant information collecting device 1 of the embodiment, the position detection device 40 is provided in order to obtain a more accurate photographing position, but the RGB camera 41 or the MS camera 46 can obtain a highly accurate photographing position GK. In this case, the position information GK of both cameras may be used. In this case, the position detection device 40 is not necessary, and the photographing position GK does not need to be changed based on the position information GN.

Further, in the embodiment, the plant information collection device 1 collects plant information based on image data, etc., and the plant growth state determination device 100 uses the collected plant information to make determinations on harvest amount, growth status, pollen amount, etc. explained.
In contrast, the plant information collection device 1 is equipped with the configuration and determination function of the plant growth state determination device 100, and the plant information collection device 1 collects plant information and uses the collected plant information to determine the plant growth state. may be determined.

Furthermore, in the described embodiment, two sets of cameras and lights tilted diagonally forward and diagonally right are arranged in the information collecting unit 4, and target plants are arranged in a grid pattern at predetermined intervals, such as in a pear orchard or vineyard. A case has been described in which the movable unit 3 moves back and forth between each row and each column where target plants are planted in an area where the target plants are planted.
On the other hand, if the information collection unit 4 is equipped with a total of four camera and lighting sets, two sets diagonally forward and diagonally to the right, and two further diagonally rear and diagonally left, the target It is also possible to move once between each row and each column of plants (one round trip for two rows and two rows).

Note that this embodiment can also be configured as follows.
(1) In the embodiment of configuration 1, an illumination means for illuminating the plant from the ground side, a photographing means for photographing the growing state of the plant illuminated by the illumination means from the ground side, and position information photographed by the photographing means. The present invention provides a plant information collection device characterized by comprising: a positional information acquisition means for acquiring a positional information; and an outputting means for outputting an image photographed by the photographing means and the acquired positional information.
(2) The embodiment of configuration 2 is characterized in that the illumination means illuminates a direction inclined in a predetermined direction with respect to a vertically upward direction, and the photographing means photographs a direction inclined in the predetermined direction. A plant information collecting device according to configuration 1 is provided.
(3) The embodiment of configuration 3 provides the plant information collecting device according to configuration 2, wherein the inclination angle in the predetermined direction is in a range of 20 degrees to 30 degrees.
(4) In the embodiment of configuration 4, the illumination means has a first illumination device and a second illumination device, and the inclined direction of the first illumination device and the inclined direction of the second illumination device are are perpendicular to each other, the photographing means has a first photographing device and a second photographing device, and the inclined direction of the first photographing device and the inclined direction of the second photographing device are perpendicular to each other. The present invention provides a plant information collection device according to configuration 2 or configuration 3, characterized in that:
(5) In the embodiment of configuration 5, any one of configurations 1 to 4, wherein the illumination means illuminates the plant with visible light, and the photographing means photographs a visible image of the plant. A plant information collection device according to configuration (1) is provided.
(6) In the embodiment of configuration 6, from configuration 1, wherein the illumination means illuminates the plant with red light and near-infrared light, and the photographing means photographs a multispectral image of the plant. A plant information collection device according to any one of configurations 4 is provided.
(7) In the embodiment of configuration 7, a base section on which the illumination means, the photographing means, the position information acquisition means, and the output means are mounted, and a moving means for moving the base section on the ground side of the plant. A plant information collection device according to any one of configurations 1 to 6, characterized in that the device comprises:
(8) In the embodiment of configuration 8, as described in configuration 7, the moving means is a tire that rolls on the ground, an automatic vehicle, a drone, or a wheel that rolls when pushed. We provide plant information collection equipment.
(9) In the embodiment of configuration 9, the feature point extraction means extracts predetermined feature points from the image photographed by the photographing means, and the image is synthesized by comparing the feature points of adjacent images with the position information. The present invention provides a plant information collection device according to any one of configurations 1 to 8, characterized in that the device includes an image synthesizing device that outputs the synthesized image, and a synthesized image output device that outputs the synthesized image.
(10) In the embodiment of configuration 10, an image information acquisition means for acquiring an image output from the plant information collection device according to any one of configurations 1 to 9 and photographing position information; a feature point extracting means for extracting feature points of the images; an image synthesizing means for synthesizing images by comparing feature points of adjacent images having the photographing position information; A plant growth state determination device is provided, comprising a growth state determination means for determining the growth state of a photographed plant.
(11) In the embodiment of configuration 11, an image information acquisition function that acquires an image output from the plant information collection device according to any one of configurations 1 to 9 and photographing position information, and a predetermined information from the acquired image are provided. a feature point extraction function that extracts feature points of images; an image synthesis function that synthesizes images by comparing feature points of adjacent images having the shooting position information; and an image synthesis function that synthesizes images from the synthesized images. To provide a plant growth state determination program for causing a computer to realize a growth state determination function for determining the growth state of a photographed plant.

Furthermore, it is also possible to configure the following embodiment as an embodiment for estimating the yield of fruit.
(21) In the embodiment of configuration 21, a visible image acquisition means for acquiring a visible image of a target fruit tree in a flowering state, and a relationship between flowers and the number of harvested fruits defined for the target fruit tree are defined. a condition acquisition means for acquiring a determination condition obtained from the above-mentioned determination condition; a flower state acquisition means for acquiring a flower condition from the visible image according to the acquired determination condition; and a fruit corresponding to the acquired flower condition according to the acquired determination condition. A harvest amount estimating device is provided, comprising: an estimating means for estimating a harvest amount; and an output means for outputting the estimated harvest amount.
(22) In the embodiment of configuration 22, the harvesting according to configuration 21, wherein the visible image acquisition means acquires a visible image photographed from the ground side with the target fruit tree illuminated after sunset. A quantity estimation device is provided.
(23) In the embodiment of configuration 23, the target fruit tree is a pear, the condition acquisition means acquires the relationship between the flower cluster and the number of harvests as the determination condition for the pear, and the flower state acquisition means acquires the relationship between the flower cluster and the number of harvests from the visible image. Yield estimation according to configuration 21 or configuration 22, characterized in that the number of flower clusters is acquired as the state of flowers, and the estimating means estimates the harvested number of pears corresponding to the acquired number of flower clusters according to the determination condition. Provide equipment.
(24) In the embodiment of configuration 24, the condition acquisition means further acquires a determination condition that sets the number of harvests of a predetermined number of flower clusters from the tips of branches to zero, the yield amount according to configuration 23. An estimation device is provided.
(25) In the embodiment of configuration 25, the target fruit tree is a fruit tree such as a pear, and the condition storage means stores a determination condition that defines the relationship between flowers and the number of harvested fruits for each fruit tree, and the condition acquisition means 23. The harvest amount estimating device according to configuration 21 or claim 22, wherein a determination condition specified in correspondence with the acquired visible image is acquired.
(26) In the embodiment of configuration 26, the yield estimation according to configuration 25, wherein the determination conditions stored in the condition storage means include a plurality of determination conditions defined for each fruit tree. Device.
(27) In the embodiment of configuration 27, a visible image acquisition function that acquires a visible image of a target fruit tree in a flowering state, and a relationship between flowers and the number of harvested fruits defined for the target fruit tree are defined. a condition acquisition function that acquires a determination condition that has been obtained, a flower state acquisition function that acquires a flower state from the visible image according to the acquired determination condition, and a fruit that corresponds to the acquired flower state according to the acquired determination condition. The present invention provides a yield estimation program for causing a computer to realize an estimation function for estimating the yield of , and an output function for outputting the estimated yield.

1 Plant information collection device 2 Base 3 Moving unit 4 Information collection unit 5 Control unit 6 Power source 40 Position detection device 41 RGB camera 42 LED light 46 MS camera 47 Red light light 48 Near-infrared light light 50 Lighting switch 51 CPU
52 ROM
53 RAM
531 Photography information 532 Visible image 533 MS image 534 Location information GN
54 Input/output section 55 Storage section 551 Plant information collection program 552 Movement area 553 Collected plant information 56 Communication control section 100 Plant growth state determination device 110 CPU
120 ROM
130 RAM
132 Visible image 133 MS image 134 Position information GN
136 Feature points 137 Composite image 138 Judgment result 140 USB terminal 150 Input section 160 Display device 170 Communication control section 180 Storage section 181 Growth state determination program 182 Collected plant information 183 Judgment condition 184 Plant growth state determination information α Tilt angle β Tilt angle D , E Male flower index F Branching point G Green area GK Photographing position GN Position information H Nodes KS1 to KS3 Floral cluster PL Tree Q Male flower density R Coefficient of determination U Number of fruits in the floral cluster RL, RR Range line S Optical axis T Harvest amount s1, s2 Photographing Interval V Vertical line W Number of target flower clusters KH Pollen production amount p1, p2 Photographing position

Claims (11)

illumination means for illuminating plants from the ground side;
Photographing means for photographing the growing state of the plants illuminated by the illumination means from the ground side;
position information acquisition means for acquiring position information photographed by the photographing means;
output means for outputting an image photographed by the photographing means and the acquired position information;
A plant information collection device comprising: The illumination means illuminates a direction inclined in a predetermined direction with respect to a vertically upward direction,
The photographing means photographs a direction inclined to the predetermined direction.
The plant information collection device according to claim 1, characterized in that: The angle of inclination in the predetermined direction is in the range of 20 degrees to 30 degrees.
3. The plant information collection device according to claim 2. The illumination means includes a first illumination device and a second illumination device, the inclined direction of the first illumination device and the inclined direction of the second illumination device are perpendicular to each other,
The photographing means has a first photographing device and a second photographing device, and the inclined direction of the first photographing device and the inclined direction of the second photographing device are perpendicular to each other.
3. The plant information collection device according to claim 2. The illumination means illuminates the plant with visible light,
The photographing means photographs a visible image of the plant.
The plant information collection device according to claim 1, characterized in that: The illumination means illuminates the plant with red light and near-infrared light,
The photographing means photographs a multispectral image of the plant.
The plant information collection device according to claim 1, characterized in that: a base portion on which the illumination means, the photographing means, the position information acquisition means, and the output means are mounted;
a moving means for moving the base on the ground side of the plant;
The plant information collection device according to claim 1, further comprising: The moving means includes tires that roll on the ground, an automatic vehicle, a drone, or wheels that roll when pushed.
8. The plant information collection device according to claim 7. feature point extraction means for extracting predetermined feature points from the image photographed by the photographing means;
image synthesis means for synthesizing images by comparing feature points of images with adjacent position information;
a composite image output means for outputting the composite image;
The plant information collection device according to claim 1, comprising: Image information acquisition means for acquiring images and photographing position information output from the plant information collection device according to any one of claims 1 to 9;
Feature point extraction means for extracting predetermined feature points from the acquired image;
image synthesis means for synthesizing images by comparing feature points of images with adjacent shooting position information;
A growing state determining means for determining the growing state of the plant photographed by the plant information collecting device from the combined image;
A plant growth state determination device comprising: an image information acquisition function that acquires an image and photographing position information output from the plant information collection device according to any one of claims 1 to 9;
a feature point extraction function that extracts predetermined feature points from the acquired image;
an image synthesis function that synthesizes images by comparing feature points of images with adjacent shooting position information;
a growth state determination function that determines the growth state of the plant photographed by the plant information collection device from the combined image;
A plant growth status determination program that allows computers to realize this.

Images