TW202107972A - Unmanned aerial system and liquid spraying method with artificial intelligence image processing technology capable of accurately controlling irrigation and spraying timing for water/liquid fertilizer/medicine solution required for crop growth - Google Patents

Abstract

This invention discloses an unmanned aerial system and a water/liquid fertilizer/medicine solution spraying method with artificial intelligence image processing technology. In which, the unmanned aerial vehicle is mounted with a spraying device and a multispectral camera. The farming area is divided into multiple blocks. Each block is set with navigation parameters. According to the spraying flight path, the unmanned aerial vehicle is controlled to fly through each block in sequence and the multispectral camera is used to capture the multispectral image for each block. The navigation parameters and multispectral images of each block are transmitted to the monitoring unit through signal transmission means. The image processing module built in the monitoring unit will perform the image processing on the multispectral images and interpret the navigation parameters for each block. The image processing module calculates a projected leaf area index and a normalized vegetation index for each block based on the multispectral images to determine the blocks to be sprayed with water/liquid fertilizer/medicine solution and the spraying amount of each block, so as to accurately control the irrigation and spraying timing of the water/liquid fertilizer/medicine solution required for crop growth. Therefore, this invention may improve the pesticide abuse issue, reduce the cost of agricultural planting, and enhance the efficiency of agricultural pest management.

Description

Water/liquid fertilizer/liquid spraying method of drone system and artificial intelligence image processing technology

The present invention relates to an unmanned aerial vehicle system and artificial intelligence image processing technology for water/liquid fertilizer/medicinal liquid spraying method, in particular to an unmanned aerial vehicle that can precisely control the spraying timing of agricultural water/liquid fertilizer/medicine liquid irrigation required for crop growth Agricultural implementation technology.

According to, the application of Unmanned Aerial System (UAS) is used to collect photogrammetry and remote sensing detection data. In recent years, it has been widely concerned by various research units, business circles and relevant government departments. Compared with satellite vehicles and manned airborne platforms, unmanned aerial systems have the advantages of low personnel operating risks, low operating costs, high efficiency and high spatial resolution image acquisition. In the field of remote sensing detection, multispectral and hyperspectral cameras are indispensable for obtaining visible light (such as red light, green light, blue light) and invisible light (such as red light edge and near-infrared light) information The sensor, which captures images of various wavelengths, can deduce more than 70 plant life indicators, which is suitable for applications in precision agriculture, plant life surveys, and disaster assessment. Therefore, the unmanned aerial system is equipped with a multi-spectral/hyperspectral camera It is indeed a high-efficiency plant monitoring technology solution.

Known precision agricultural application technologies often use drone vehicles and are equipped with multispectral cameras to record multispectral images of the growth status of large areas of crops to identify which areas of the farming area require special attention. Therefore, multispectral cameras do It has become the most suitable payload for UAV vehicles. However, the multi-spectral camera used is hyperspectral at the megapixel level. The (hyperspectral) camera is expensive (over NT$2 million) and heavy (over 1.5kg). It also requires the drone vehicle to hover for more than 7 seconds to capture images, resulting in image capture. Inconvenience and troubles caused by excessive access and cost expenditures.

In order to solve this deficiency, in recent years, European and American manufacturers have successively introduced a number of relatively inexpensive multi-spectral cameras suitable for drones. Generally, they have 4 to 6 frequency bands and weigh about 150g. They are as convenient and fast as RGB cameras. With practicality and price advantage, it is the best payload for drones used in delicate agriculture. For example, the "Sequoia" multi-spectral camera can have RGB and near-infrared lenses, which can cover four different spectral bands, including: green light (wavelength 500nm), red light (wavelength 660nm), red light edge (wavelength 735nm) and near Infrared light (wavelength 790mn). Multispectral images use different objects to have different reflection characteristics for each band spectrum, which is very convenient for resource investigation and environmental exploration.

According to investigations, the patents for the use of drone vehicles for agricultural purposes are shown in the US Invention Publication No. US2017/0127606, US Invention No. US9745060, and World Intellectual Property Organization No. WO2018/000399. These patents are based on the application of remote-controlled drone vehicles for agricultural irrigation, fertilization, and pesticide application. The overall technical characteristics of precision agriculture applications analyzed by multi-spectral imaging software have not been disclosed, so these patents cannot be accurate The timing of irrigation and spraying of water/liquid fertilizer/chemical liquid required for land control of crop growth cannot effectively improve the problem of pesticide abuse, and cannot reduce agricultural planting costs and improve the efficiency of agricultural pest management.

In view of this, the above-mentioned conventional precision agriculture application technology and the above-mentioned patents are indeed not perfect, and there is still a need for improvement, and based on the urgent needs of related industries, the inventors have made continuous efforts Under the research and development, a set of the invention is finally developed which is different from the above-mentioned conventional technology and the previously disclosed patent.

The first objective of the present invention is to provide a water/liquid fertilizer/liquid spraying method based on UAV system and artificial intelligence image processing technology, which can precisely control the irrigation and spraying timing of the water/liquid fertilizer/liquid required for crop growth. Therefore, it has many characteristics such as improving the problem of pesticide abuse, reducing the cost of agricultural planting, and improving the efficiency of agricultural pest management. The technical means to achieve the aforementioned first objective is to install spraying devices and multi-spectral cameras on the drone vehicle. Divide the farming area into multiple blocks, and set sailing parameters for each block. According to the spraying flight path, the UAV vehicle is controlled to fly through each block in sequence, and the multispectral camera is used to capture the multispectral image in each block. The navigation parameters and multi-spectral images of each block are transmitted to the monitoring unit through signal transmission means. The multi-spectral image is processed by the built-in image processing module of the monitoring unit, and the navigation parameters of each block are interpreted. The image processing module calculates the projected leaf area index and return of each block based on the multi-spectral image. A vegetation index is used to determine the area where the water/liquid fertilizer/chemical liquid needs to be sprayed and the amount of spraying in the area.

The second objective of the present invention is to provide a water/liquid fertilizer/medicine spraying method for an unmanned aerial vehicle system with an inter-image capturing function and artificial intelligence image processing technology, which is mainly used with a divergent short-wave light source on the unmanned aerial vehicle vehicle. For the purpose of detecting crop adversity, it is mainly to use dynamic fluorescent indicators as the numerical indicators after the drone takes the phase, as the estimation of the physiological state of the crops after the adversity. The technical means to achieve the aforementioned second objective is to install a spray device and a multi-spectral camera on the drone vehicle. Divide the farming area into multiple blocks, and set sailing parameters for each block. According to the spraying flight path, the UAV vehicle is controlled to fly through each block in sequence, and the multispectral camera is used to capture the multispectral image in each block. The navigation parameters and multi-spectral images of each block are transmitted to the monitoring unit through signal transmission means. The multi-spectral image is processed by the built-in image processing module of the monitoring unit, and the navigation parameters of each block are interpreted. The image processing module calculates the projected leaf area index and return of each block based on the multi-spectral image. A vegetation index, and determine Determine the area to be sprayed with water/liquid fertilizer/liquid and the amount of spraying in the area. It further includes a night image capturing step, which is to install a short-wave light source device on the UAV vehicle to control the UAV vehicle to sequentially fly over each block of the farming area at night, and every One block performs multi-spectral image capture; when the UAV vehicle arrives in one of the blocks, the short-wave light source device is turned on to emit the band-wave light source for the crop in the block, and the multi-spectral camera is activated. 25 to 35 seconds to capture the multi-spectral image of the block; the image processing module converts the captured multi-spectral image into a fluorescent index, and draws the fluorescent index value into a contour map, In order to obtain the fluorescence index intensity and difference characteristics of the location of the block, the physiological state of the crop is estimated after adversity, so as to be the basis for adjusting the spray amount of the water/liquid fertilizer/medicine.

The third object of the present invention is to provide an unmanned aerial vehicle system and artificial intelligence image processing technology that enables the UAV vehicle to perform agricultural fertilizer spraying schedule setting tasks to obtain navigation parameters of each area as the basis for planning the spraying navigation path. Liquid/liquid fertilizer/liquid spraying method. The technical means to achieve the aforementioned third objective is to install a spray device and a multi-spectral camera on the drone vehicle. Divide the farming area into multiple blocks, and set sailing parameters for each block. According to the spraying flight path, the UAV vehicle is controlled to fly through each block in sequence, and the multispectral camera is used to capture the multispectral image in each block. The navigation parameters and multi-spectral images of each block are transmitted to the monitoring unit through signal transmission means. The multi-spectral image is processed by the built-in image processing module of the monitoring unit, and the navigation parameters of each block are interpreted. The image processing module calculates the projected leaf area index and return of each block based on the multi-spectral image. A vegetation index is used to determine the area where the water/liquid fertilizer/chemical liquid needs to be sprayed and the amount of spraying in the area. It further includes a flight control unit and a spraying range setting module provided on the drone vehicle. The spraying range setting module and the flight control unit are in signal communication via the signal transmission means; the spraying range setting module can For setting to generate at least one control signal, the at least one control signal is transmitted to the flight via the signal transmission means The control unit controls the UAV vehicle to execute the flight path setting step according to the flight path. When the flight path setting step is executed, the UAV vehicle flies over according to a predetermined altitude and a predetermined speed in the sequence of the path The center coordinate position above each block of the farming area, and the navigation parameters of each center coordinate position are set or recorded. The navigation parameters include serial number parameters, speed parameters, altitude parameters, center coordinate position parameters and arrival time Parameters, when each of the blocks has completed the setting or recording of the sailing parameters, the spraying voyage setting module generates the flight path, which is transmitted to the monitoring unit by the signal transmission means, and interpreted by the image processing module The flight path is then corrected to the spraying flight path according to the projected leaf area index and the normalized vegetation index, so that the flight control unit controls the UAV vehicle to perform corresponding flight control and control according to the spraying flight path. Actions such as spraying control of the water/liquid fertilizer/chemical liquid.

10‧‧‧UAV Vehicle

11‧‧‧Flight Control Unit

20‧‧‧Signal transmission method

21‧‧‧Second signal transmission method

30‧‧‧Monitoring unit

31‧‧‧Image Processing Module

40‧‧‧Spray device

50‧‧‧Multispectral Camera

60‧‧‧Spraying Range Setting Module

70‧‧‧Shortwave light source device

a‧‧‧Agricultural area

a1‧‧‧block

A1‧‧‧Plant area

A2‧‧‧Total area of blade projection

O‧‧‧origin

dn‧‧‧Spray flight path

Figure 1 is a schematic diagram of the operation and implementation of the unmanned aerial vehicle carrier of the present invention.

Fig. 2 is a schematic diagram of the implementation of the spraying device and the multispectral camera on the unmanned aerial vehicle carrier of the present invention.

Fig. 3 is a schematic diagram showing the comparison of the spraying opening and closing control of the spraying device of the present invention and the area to be irrigated.

Figure 4 is a schematic diagram of the implementation of the drone carrier of the present invention flying along the spraying flight path.

Figure 5 is a schematic diagram of the implementation of the puzzle of the RGB spectral image of the present invention.

Fig. 6 is a schematic diagram of a puzzle of multi-spectral images of green light, red light, red light edge, and near-infrared light from left to right of the present invention.

Figure 7 (a) is the NIR image puzzle of the present invention; (b) is the DI image puzzle of the present invention; (c) is the SR image puzzle of the present invention; (d) is the GNDI image puzzle of the present invention; (e) is the MSAVI image of the present invention Schematic diagram of the puzzle implementation.

Fig. 8 is a schematic diagram of the control growth curve of the projected leaf area index and the growth days of the seedlings of the present invention.

Fig. 9 is a schematic diagram showing the growth state of the normalized vegetation index of the present invention.

Fig. 10 is a schematic diagram of comparative analysis of the growth state of the seedlings of the present invention and several environmental factors, NDVI and PLAI indexes.

Fig. 11 is a schematic diagram of the comparison curve between the fluorescence intensity and the fluorescence emission time of the present invention.

Fig. 12 is a schematic diagram of the comparison curve between the water potential and the fluorescence index distance of the present invention.

Figure 13 is a schematic diagram of the comparison curve between the fluorescence index and the water content of the present invention.

Fig. 14 is a schematic diagram of the comparison curve between the moisture potential and the fluorescence reduction rate of the present invention.

Fig. 15 is a schematic diagram of the comparison of the height of the projected leaf area index of the present invention; (a) is a schematic diagram of the distribution where the projected leaf area index is higher than 0.85; (b) is a schematic diagram of the distribution where the projected leaf area index is lower than 0.85.

In order for your reviewer to further understand the overall technical features of the present invention and the technical means to achieve the purpose of the invention, specific examples and diagrams are used to illustrate in detail: please refer to Figures 1 to 4 for the purpose of achieving the invention. The first specific embodiment of the first item includes the following steps:

(a) Preparation steps: provide an unmanned aerial vehicle 10, a signal transmission means 20 and a monitoring unit 30. Among them, the unmanned aerial vehicle 10 is equipped with a spraying device 40 capable of spraying water/liquid fertilizer/medicinal liquid and a multi-spectral camera 50. The signal transmission means 20 may be a wireless communication module such as RF, UHF, or VHF separately arranged on the monitoring unit 30 and the UAV vehicle 10.

(b) Flight path setting step: A farming area a is divided into a plurality of adjacent blocks a1, and each block a1 is set with a navigation parameter including a central coordinate position P for setting a flight path. Specifically, in this step, the central coordinate positions P of each block a1 from an origin O to the farming area a are sequentially connected in series to form a flight path.

(c) Image capturing step: control the drone vehicle 10 to fly through each block a1 of the farming area a in sequence according to the flight path, and use the multispectral camera 50 to capture multispectral images in each block a1 .

(d) Information acquisition step: the navigation parameters and multi-spectral images of each block a1 are transmitted to the monitoring unit 30 through the signal transmission means 20.

(e) Image processing step: use an image processing module 31 built in the monitoring unit 30 to perform image processing on the multi-spectral images, and sequentially apply each multi-spectral image to a projection leaf area index formula and a normalized vegetation The index formula is used to calculate the projected leaf area index (PLAI) and normalized vegetation index (NDVI) of each block a1, and then re-correct the flight path to a spray flight path dn according to the projected leaf area index and normalized vegetation index ; Among them, the projected leaf area index formula is the total projected leaf area A2/plant area A1; the normalized vegetation index formula is (near infrared light reflection ρ _NIR -red light reflection ρ _RED )/(near infrared The amount of light reflection ρ _NIR + the amount of red light reflection ρ _RED ).

(f) Spraying execution steps: Control the drone vehicle 10 to fly through the farming area in sequence according to the spraying flight path dn a. The area a1 that needs to be sprayed with water/liquid fertilizer/chemical liquid, and based on the projected leaf area index and normalization The vegetation index determines the amount of spraying in the area a1 that needs to be sprayed.

In a specific operation embodiment, when the projected leaf area index of the block a1 is higher than 0.85, the spraying device 40 is activated to spray the water/liquid fertilizer/chemical solution to the block a1, and at the same time, when the area of the block a1 When the normalized vegetation index reaches 0.75, set the block a1 as the standard water/liquid fertilizer/chemical spray amount; when the normalized vegetation index is higher than 0.75, increase the water/liquid fertilizer/of the block a1 The spray amount of liquid medicine; when the normalized vegetation index is lower than 0.75, the spray amount of water/liquid fertilizer/liquid liquid in the block a1 is reduced.

Please refer to FIGS. 1 to 4, in order to achieve the second specific embodiment of the second object of the invention, this embodiment includes the overall technical features of the first specific embodiment described above, and is further used in the above-mentioned image capturing step. Including a night image capture step, which is installed on UAV vehicle 10 A short-wave light source device 70 is used to control the drone vehicle 10 to sequentially fly over each block a1 of the farming area a at night, and perform multi-spectral image capture in each block a1. When the drone vehicle 10 reaches one of the blocks a1, the short-wave light source device 70 is turned on to emit a band-wave light source to the crop in the block a1, and the multi-spectral camera 50 is activated for about 25 to 35 seconds (preferably 30 seconds) , To capture the multi-spectral image of block a1. The image processing module 31 converts the captured multi-spectral image into a fluorescent index, and draws the fluorescent index value into a contour map to obtain the fluorescent index intensity and difference characteristics at the position P of the block a1, and After the physiological state of the crop is subjected to adversity, the growth status is estimated, which is used as the basis for adjusting the amount of water/liquid fertilizer/medicine spray.

Specifically, the above estimation refers to water stress, nutrient stress, temperature stress, salinity stress, bacterial infection stress, or pest stress. More specifically, the above-mentioned water liquid/liquid fertilizer/medicinal liquid may be one of irrigation water, salt water, liquid fertilizer, or liquid pesticide. When the growth condition of block a1 is estimated to be moisture or temperature adversity, increase the amount of irrigation water sprayed in this block a1. When the growth status of the block a1 is estimated to be nutritional adversity, the amount of liquid fertilizer sprayed in the block a1 is increased. When the growth condition of the block a1 is estimated to be temperature adversity, the irrigation water spraying amount of the block a1 is increased. When the growth condition of the block a1 is estimated as salinity adversity, the amount of salt water sprayed in the block a1 is increased. When the growth status of the block a1 is estimated to be bacterial infection or pest adversity, the amount of liquid pesticide spraying in the block a1 is increased.

Please refer to Figures 1 to 4, in order to achieve the third specific embodiment of the third item of the invention, this embodiment not only includes the overall technical features of the above-mentioned first specific embodiment, but also includes an unmanned aerial vehicle. A flight control unit 11 with 10 and a spraying range setting module 60 (such as a remote remote controller; but not limited to this). The spraying range setting module 60 and the flight control unit 11 are in signal communication via the signal transmission means 20. The spraying range setting module 60 can be set to generate at least one control signal. The at least one control signal is transmitted to the flight control unit 11 via the signal transmission means 20 to control the drone vehicle 10 to execute the flight path setting step according to the flight path. , Execute flight path design When determining the steps, the UAV vehicle 10 flies over the center coordinate position P above each block a1 of the farming area a according to a predetermined height and a predetermined speed and in the order of the path, and sets or records each center coordinate position P Navigation parameters, which include serial number parameters, speed parameters (the height can be confirmed through the speed sensor), altitude parameters (the height can be confirmed through the height sensor), the center coordinate position parameters (the position can be confirmed through the GPS), and arrival Time parameters. When each block a1 has completed the setting or recording of the sailing parameters, the spraying voyage setting module 60 generates the above-mentioned flight path, which is transmitted to the monitoring unit 30 by a second signal transmission means 21, and then is transmitted to the monitoring unit 30 through the image processing module 31 After interpreting the flight path, it is corrected to the above-mentioned spraying flight path dn according to the projected leaf area index and the normalized vegetation index, and then transmitted to the flight control unit 11 via the signal transmission means 20, so that the flight control unit 11 can calculate according to the spraying flight path dn Control the UAV vehicle 10 to perform corresponding flight control and spray control of water/liquid fertilizer/medicine liquid and other actions.

In a specific operating embodiment, before the drone vehicle 10 is flying, the overlap rate of the left and right and the front and back of the adjacent multispectral images of the multispectral camera 50 can be preset, and each multispectral image captured by the multispectral camera 50 can be set in advance. The images are all recorded with GPS coordinate position, height, and shooting posture. Then, through the built-in puzzle software of the image processing module 31, a complete and larger multi-spectral image is formed according to different spectra. Figure 5 shows the result of the puzzle of the RGB spectrum. Figure 6 shows a mosaic of multi-spectral images such as green light, red light, red light edge, and near-infrared light from left to right. In addition, the vegetation index (Vegetation index) of each point can also be calculated from the above photos of different spectra using the following formula. Figure 7(a) is the result of splicing near infrared light (NIR) spectrum images of rice aerial photography. By analyzing the image values obtained in different wavebands, many studies have further proposed the vegetation index formula, as shown in Table 1 below, hoping to grasp the crop yield more accurately. For example, FIG. 7(b) shows the calculated difference index DI (Difference Index) Vegetation Index image using near infrared spectroscopy (NIR) image and green light (Green) image. Will crop In a multispectral image, the intensity of infrared light minus the intensity of visible light is called DVI, Difference Vegetation Index. If you divide by the infrared light intensity plus the visible light intensity, it is called NDVI. (The reason for dividing by this value is because the intensity of light irradiated by each place is not necessarily the same, so divide by this value to be standardized) To sum up, the formula is NDVI=(IR-VIS)/(IR+VIS) IR is infrared Light VIS is visible light; in addition, Fig. 7(c) is an SR image puzzle; Fig. 7(d) is a GNDI image puzzle; Fig. 7(e) is an implementation schematic diagram of an MSAVI image puzzle.

Then, calculate the following two crop indicators:

(1) Projected leaf area index (PLAI): Leaf is the main organ of plants for photosynthesis, respiration, transpiration, carbon cycle, and precipitation interception, and leaf area is the indicator of crop production in crop physiology and agricultural research. The most effective measurement item of potential. The change and size of leaf area can show the degree of crop growth and the ability to intercept light energy, etc., which are important traits for crop growth analysis. Its calculation formula is the projected leaf area index. Fig. 8 shows the contrast growth curve of the projected leaf area index and seedling growth days of the present invention.

When photographing, the multi-spectral camera 50 is as far as possible to take pictures from the top of the leaf at 90 degrees, thereby reducing the leaf area estimation error caused by image distortion.

(2) Normalized difference vegetation index (NDVI): used to analyze the information obtained from remote sensing observations. NDVI is usually calculated using satellite remote sensing data to assess the growth status of green vegetation in the target area. The calculation method is to use the reflection amount of red light ρ _RED and the reflection amount of near-infrared light ρ _NIR , which can show information such as plant growth, ecosystem vitality and productivity. The larger the value, the more plants grow. The formula is as follows:

The value of NDVI is between -1 and 1. When ρ _RED =0, there is a maximum value of 1; on the contrary, when ρ _NIR =0, there is a minimum value of -1. At the same time, the following three environmental conditions of each crop are measured, including temperature, relative humidity, and lighting conditions to improve the uniformity of these environmental conditions, and to understand the growth conditions of plants, and implement site-specific cultivation (site-specific cultivation) , This is the spirit of precision agriculture. Finally, the PLAI index and the NDVI index value are used to determine the amount of crop irrigation. The irrigation switch has two states, the state when watering is turned on is 1, and the state when no watering is turned off is 0. When the standard value of PLAI is higher than 0.85, the present invention sets the watering switch status to 1; otherwise, when the standard value of PLAI is lower than 0.85, the present invention sets the watering switch status to 0, and the NDVI value (0.75) is added to evaluate the watering In the standard of water level. As shown in the figure below, establish an xy coordinate system (divided into continuous. Some fixed-size small blocks) within the crop area, and then calculate the PLAI and NDVI values of each point, and determine the irrigation valve of each coordinate point according to the set threshold switch status. Figure 9 shows a schematic representation of the growth state of the normalized vegetation index of the present invention. Figure 10 shows the comparison analysis of the growth state of the seedlings of the present invention and several environmental factors, NDVI and PLAI index.

If the irrigation switch is linear, the opening amount u(t) of the switch can be set from 0 to 1. 0 means that the switch is not turned on, and 1 means that the switch is fully open. The present invention u(x,y) is determined by the PLAI index value and the NDVI index value of the crop at the coordinate point. One of the embodiments is: u _w (x,y)=MAX{PLAI(x,y),NDVI( x,y)}

Among them, u(x,y) can be _{the valve control amount of spraying water u w} (x, y) or fertilizing u _f (x, y) or pesticide u _p (x, y) on the drone.

The decision of u(x,y) can be u _w (x,y)=w(PLAI(x,y), ρ _blue (x,y), ρ _green (x,y), ρ _red (x,y) ), ρ _{red edge} (x,y), ρ _NIR (x,y))

u _f (x,y)=f(PLAI(x,y), ρ _blue (x,y), ρ _green (x,y), ρ _red (x,y), ρ _{red edge} (x,y), ρ _NIR (x,y))

u _P (x,y)=P(PLAI(x,y), ρ _blue (x,y), ρ _green (x,y), ρ _red (x,y), ρ _{red edge} (x,y), ρ _NIR (x,y))

In addition, the present invention can use the drone carrier 10 to mount a multi-spectral camera 50 and match it with a lightweight shortwave light source device 70 capable of diverging shortwave light sources on the drone carrier 10. For the detection of crop adversity, The use of dynamic fluorescent indicators as the numerical indicators after the drone takes the phase can be used to estimate the physiological state of crops after adversity. The estimate after adversity can be water adversity, nutritional adversity, temperature adversity, or It is salinity adversity, these adversities will be different in fluorescent performance, which can be used as the basis for analysis and judgment after the drone takes the fluorescent images at night. And increase the shooting function of the drone, taking night shooting as the difference and distinguishing from other drone vehicles 10.

Before shooting at night, dark processing is required; that is, before the multi-spectral image is taken, the crop cannot be illuminated by other light sources, so as not to affect its physiological response. After the drone vehicle 10 performs fixed-point shooting, it must be at a fixed-point Turn on the short-wave light source device 70 on the upper side, as shown in FIG. 1, and use the multi-spectral camera 50 for image capture. The multi-spectral camera 50 on the drone vehicle 10 must be able to continuously capture at a fixed point for thirty seconds to obtain sufficient The amount of fluorescence, and can calculate the dynamic fluorescence index, Therefore, in the field of image capture, it is necessary to be able to perform precise location positioning of latitude and longitude in order to capture the effective location of multi-spectral image information, for the embodiment of water adversity.

In addition, for the diagnosis of diseases and insect pests, based on fluorescent images and the multispectral image information of the crop itself, crops suffering from diseases and insect pests will cause damage to the chloroplasts in the leaves, which is manifested in the chlorophyll fluorescence. In the state caused by the disease, it will most likely cause damage. Leaf atrophy or diseased spots directly affect the chloroplasts in plant leaves. Therefore, when you are infected by bacteria, molds, fungi or viruses, the local leaf fluorescence response will have different fluorescence responses, such as Bacterial lesions will cause the intensity of fluorescence in the infected area to be stronger, and the intensity of fluorescence is different from that of normal leaf tissue.

For the diagnosis of pests, if there are nymphs, adults, or eggs on the back of leaves or curled leaves, there will also be different changes in multispectral images and fluorescent images. These changes can be used for data diagnosis of pests. Different image intensity signals and regional judgments can provide the diagnosis of blade damage.

Then, please refer to Figure 11, where RF: refers to normal watering, and EA is one-day water-off treatment, EB is two-day water-off treatment, EC is three-day water-off point, and ED is four-day water-off treatment. Treatment; observe the physical state after water treatment and compare it with the fluorescence image curve. The EA and EB groups have a chance to recover before the permanent withering point, but the curve performance of EC and ED is that the crop lacks water A state that is difficult to recover.

Please refer to Figure 12 again, which is the curve of fluorescence index and water potential, which is used as the basis and scientific evidence for the conversion of fluorescence image to water management. This regression curve can be used as a scientific indicator basis for the online UAV fluorescent imaging system. Please refer to Figure 13. The data used in traditional fluorescence detection is RFD, which is defined as (Fm/Fs)-1, which is the ratio of the maximum value of the fluorescence curve to the steady-state value minus one, so it is transmitted through traditional fluorescence The detected data is used for regression analysis with the moisture content of the leaves, and the result will have a large standard error. Therefore, it is not recommended to be used for online drones. Use of fluorescent imaging system. Please refer to Figure 14. The regression analysis of the fluorescent indicators of traditional plant physiology research and the water potential of crops is relatively divergent, and there is no convergence phenomenon, and it is not suitable for using on-line UAV fluorescent imaging system.

Please continue to refer to Figure 15(a), which is a schematic diagram of the area distribution with a projected leaf area index higher than 0.85, and Figure 15(b) is a schematic diagram of the area distribution with a projected leaf area index lower than 0.85.

The above is only a more feasible embodiment of the present invention, and is not intended to limit the patent scope of the present invention. Any equivalent implementation of other changes based on the content, characteristics and spirit of the following claims is mentioned. All should be included in the scope of the patent of the present invention. The structural features of the invention specifically defined in the claim are not found in similar articles, and are practical and progressive. They have already met the requirements of a patent for invention. The application is filed in accordance with the law. I would like to request that the Bureau of Junction approve the patent in accordance with the law to protect this The legitimate rights and interests of the applicant.

10‧‧‧UAV Vehicle

11‧‧‧Flight Control Unit

20‧‧‧Signal transmission method

21‧‧‧Second signal transmission method

30‧‧‧Monitoring unit

31‧‧‧Image Processing Module

40‧‧‧Spray device

50‧‧‧Multispectral Camera

60‧‧‧Spraying Range Setting Module

70‧‧‧Shortwave light source device

Claims (8)

An unmanned aerial vehicle system and artificial intelligence image processing technology water/liquid fertilizer/medical liquid spraying method, which includes: (a) preparation steps: providing an unmanned aerial vehicle, a signal transmission means and a monitoring unit; wherein, in the The UAV vehicle installation includes a spraying device for spraying water/liquid fertilizer/chemical liquid and a multi-spectral camera; (b) Flight path setting steps: dividing a farming area into a plurality of adjacent blocks, each A navigation parameter including a center coordinate position is set in the block to set a flight path; (c) Image capture step: control the UAV vehicle to sequentially fly through each of the farming areas according to the flight path One block, and the multi-spectral camera is used to capture multi-spectral images in each block; (d) Information acquisition step: the navigation parameter and the multi-spectral image of each block are transmitted through the signal transmission means The image is transmitted to the monitoring unit; (e) Image processing step: image processing is performed on the multi-spectral image with an image processing module built in the monitoring unit, and each of the multi-spectral images is sequentially based on a projection leaf area The index formula and a normalized vegetation index formula are used to calculate the projected leaf area index (PLAI) and normalized vegetation index (NDVI) of each block, and then based on the projected leaf area index and the normalized vegetation index Re-correct the flight path as a spray flight path; where the projected leaf area index formula is the total projected leaf area/plant area; the normalized vegetation index formula is (near infrared light reflection ρ _NIR -red Light reflection quantity ρ _RED )/(near infrared light reflection quantity ρ _NIR + red light reflection quantity ρ _RED ); and (f) spraying execution steps: according to the spraying flight path to control the UAV vehicle to fly over the The farming area needs to spray the water/liquid fertilizer/chemical liquid in the area, and the amount of spraying in the area needs to be sprayed according to the projected leaf area index and the normalized vegetation index. The water/liquid fertilizer/liquid spraying method of unmanned aerial vehicle system and artificial intelligence image processing technology according to claim 1, wherein, in the flight path setting step, an origin is set to each of the farming areas The central coordinate positions of the block are sequentially connected in series to form the flight path. The water/liquid fertilizer/liquid spraying method of UAV system and artificial intelligence image processing technology described in claim 1, wherein when the projected leaf area index of the block is higher than 0.85, the spraying device will be activated The water/liquid fertilizer/chemical solution is sprayed to the block; when the normalized vegetation index of the block reaches 0.75, it is set as the standard water/liquid fertilizer/chemical spray amount of the block; when When the normalized vegetation index is higher than 0.75, increase the spray amount of the water/liquid fertilizer/chemical solution in the block; when the normalized vegetation index is lower than 0.75, reduce the water/liquid/medicine solution in the block The amount of liquid fertilizer/liquid sprayed. The water/liquid fertilizer/liquid spraying method of the drone system and artificial intelligence image processing technology according to claim 1, wherein the image capturing step further includes a night image capturing step, which is attached to the drone The vehicle is equipped with a short-wave light source device to control the UAV vehicle to fly through each block of the farming area in sequence at night, and perform multi-spectral image capture in each block; when the UAV When the vehicle reaches one of the blocks, the short-wave light source device is turned on to emit a band-wave light source for the crop in the block, and the multi-spectral camera is activated for about 25 to 35 seconds to capture the multi-spectrum of the block Image; the image processing module converts the captured multi-spectral image into a fluorescent index, and draws the fluorescent index value into a contour map to obtain the intensity and difference characteristics of the fluorescent index at the location of the block , To estimate the growth status after the physiological state of the crop is subjected to adversity, which is used as the basis for adjusting the spray amount of the water/liquid fertilizer/medicine. The water/liquid fertilizer/liquid spraying method of unmanned aerial vehicle system and artificial intelligence image processing technology as described in claim 4, wherein the estimation of the growth status is selected from the group consisting of water adversity, nutritional adversity, temperature adversity, salinity adversity, Bacterial infection is one of adversity and pest adversity. The water/liquid fertilizer/liquid spraying method of UAV system and artificial intelligence image processing technology according to claim 5, wherein the water/liquid fertilizer/liquid system is selected from irrigation water, salt water, liquid fertilizer and liquid pesticide When the growth condition of the block is estimated to be water or temperature adversity, increase the amount of irrigation water sprayed in the block; when the growth condition of the block is estimated to be nutritional adversity, increase the area The amount of liquid fertilizer sprayed on the block; when the growth condition of the block is estimated to be temperature adversity, the amount of irrigation water sprayed in the block is increased; when the growth condition of the block is estimated to be salinity adversity, then Increase the amount of salt water sprayed in the block; when the growth state of the block is estimated to be water adversity, increase the amount of irrigation water sprayed in the block; the growth state of the block is estimated to be bacterial infection or pests In adversity, increase the amount of the liquid pesticide sprayed in the block. The water/liquid fertilizer/liquid spraying method of UAV system and artificial intelligence image processing technology as described in claim 1, which further includes a flight control unit and a spraying range setting module provided on the UAV vehicle, The spraying range setting module and the flight control unit are in signal communication via the signal transmission means; the spraying range setting module can be set to generate at least one control signal, and the at least one control signal is transmitted to the flight via the signal transmission means The control unit controls the UAV vehicle to execute the flight path setting step according to the flight path. When the flight path setting step is executed, the UAV vehicle flies over according to a predetermined altitude and a predetermined speed in the sequence of the path The center coordinate position above each block of the farming area, and the navigation parameters of each center coordinate position are set or recorded. The navigation parameters include serial number parameters, speed parameters, altitude parameters, center coordinate position parameters and arrival time parameter. The water/liquid fertilizer/medicine spraying method of the UAV system and artificial intelligence image processing technology described in claim 7, wherein when each of the blocks has completed the setting or recording of the sailing parameters, the spraying voyage setting mode The group generates the flight path and transmits it by a second signal transmission means. It is input to the monitoring unit, the flight path is interpreted by the image processing module and then corrected to the spraying flight path according to the projected leaf area index and the normalized vegetation index, so that the flight control unit is based on the spraying flight path And control the UAV vehicle to do the corresponding flight control and the spraying control of the liquid/liquid fertilizer/medicine liquid and other actions.

Images