# US 2023/0410280 A1 · Method and system for monitoring rice bacterial blight in field based on multi-source data

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0410280 A1
- Document ID: US20230410280A1
- Status: Published application
- Kind: A1
- Application number: 18/157,972
- Filed: 2023-01-23
- Published: 2023-12-21
- Earliest claimed: 2022-05-31
- Assignees: Zhejiang University
- Applicants: Zhejiang University
- Inventors: Xuping FENG; Xiulin BAI; Yong HE; Jinnuo ZHANG; Mingzhu TAO; Qingguan WU
- Web page: https://patentgrep.com/patent/US20230410280A1

## CPC

- G06T 7/0004: {Industrial image inspection}
- A01G 22/22: Rice
- A01P 1/00: Disinfectants; Antimicrobial compounds or mixtures thereof
- B64U 20/87: Mounting of imaging devices, e.g. mounting of gimbals
- G06T 3/4038: Image mosaicing, e.g. composing plane images from plane sub-images
- B64U 2101/30: for imaging, photography or videography
- B64U 2101/40: for agriculture or forestry operations
- G06T 2207/10036: Multispectral image; Hyperspectral image
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30128: Food products
- G06T 2207/30188: Vegetation; Agriculture
- Y02A 40/10: in agriculture
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- A01G: HORTICULTURE; CULTIVATION OF VEGETABLES, FLOWERS, RICE, FRUIT, VINES, HOPS OR SEAWEED; FORESTRY; WATERING (picking of fruits, vegetables, hops or the like A01D46/00; propagating unicellular algae C12N1/12)
- A01P: BIOCIDAL, PEST REPELLANT, PEST ATTRACTANT OR PLANT GROWTH REGULATORY ACTIVITY OF CHEMICAL COMPOUNDS OR PREPARATIONS
- B64U: UNMANNED AERIAL VEHICLES [UAV]; EQUIPMENT THEREFOR
- Y02A: TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE

## Abstract

A method for monitoring rice bacterial blight includes: obtaining a multi-spectral image, severities of the rice bacterial blight, and accumulated temperature data of a rice field at different growth stages; obtaining resistance of rice varieties to the bacterial blight; extracting a mean canopy spectral reflectance of each plot in the rice field; conducting regression of the severity of the rice bacterial blight using a convolutional neural network based on the mean canopy spectral reflectance and the severity of the rice bacterial blight, and outputting a depth spectrum feature; training a disease severity regression model with the accumulated temperature data, the depth spectrum feature, and the resistance to the bacterial blight for each plot as an input and the corresponding severity as an output; and monitoring a severity of the rice bacterial blight in a to-be-monitored rice field using the disease severity regression model.

## Claims

### Claim 1

A method for monitoring rice bacterial blight in a field based on multi-source data, comprising:
obtaining a multi-spectral image of a rice field at different growth stages;
obtaining severities of the rice bacterial blight and accumulated temperature data in the rice field at different growth stages;
obtaining resistance of rice varieties planted in the rice field to the bacterial blight;
extracting a mean canopy spectral reflectance of each plot in the rice field based on the multi-spectral image, wherein the rice field is divided into a plurality of plots;
conducting regression of the severity of the rice bacterial blight using a convolutional neural network based on the mean canopy spectral reflectance and the severity of the rice bacterial blight for each plot, and outputting a depth spectrum feature;
training a disease severity regression model with the accumulated temperature data, the depth spectrum feature, and the resistance to the bacterial blight for each plot as an input and the corresponding severity of the rice bacterial blight as an output to obtain a well-trained disease severity regression model; and
inputting accumulated temperature data, a depth spectrum feature and resistance to the bacterial blight corresponding to each plot of a to-be-monitored rice field into the well-trained disease severity regression model to output a level of severity of the rice bacterial blight in the to-be-monitored rice field.

### Claim 2 (depends on 1)

The method according to claim 1, further comprising:
outputting an optimal band combination using an attention mechanism module in the convolutional neural network, wherein the optimal band combination is a combination of wavelengths corresponding to first N terms in a sequence obtained by sorting wavelength weights outputted by the attention mechanism module from large to small;
constructing a monitoring spectral index (SI) of the rice bacterial blight according to the optimal band combination; and
obtaining a monitoring SI of the rice bacterial blight in the to-be-monitored rice field, and mapping the monitoring SI of the rice bacterial blight in the to-be-monitored rice field into the multi-spectral image to obtain a visual image of the severity of the rice bacterial blight in the to-be-monitored rice field.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the obtaining multi-spectral image of a rice field at different growth stages specifically comprises:
obtaining remote sensing images of the rice field at different growth stages by an unmanned aerial vehicle (UAV) equipped with a multi-spectral camera; and
stitching a plurality of the remote sensing images to obtain the multi-spectral image of the rice field as a whole.

### Claim 4 (depends on 1)

The method according to claim 1, wherein the obtaining severities of the rice bacterial blight and accumulated temperature data in the rice field at different growth stages comprises:
determining the severity of the rice bacterial blight in each plot according to a proportion of an incidence area of the rice bacterial blight in a total area of each plot; and
recording maximum and minimum temperatures of each plot on a day when the multi-spectral image is obtained, and calculating the accumulated temperature data of each plot through an accumulated temperature formula.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the extracting a mean canopy spectral reflectance of each plot in the rice field based on the multi-spectral image comprises:
removing a background of the multi-spectral image of the rice field using a threshold segmentation method; and
extracting the mean canopy spectral reflectance of each plot from the multi-spectral image with the background removed.

### Claim 6 (depends on 1)

The method according to claim 1, wherein the disease severity regression model is a partial least square regression (PLSR) model.

### Claim 7

A system for monitoring rice bacterial blight in a field based on multi-source data, comprising:
a multi-spectral image obtaining module, configured to obtain a multi-spectral image of a rice field at different growth stages;
a bacterial blight severity and accumulated temperature data obtaining module, configured to obtain severities of the rice bacterial blight and accumulated temperature data in the rice field at different growth stages;
a bacterial blight resistance of rice varieties obtaining module, configured to obtain resistance of the rice varieties planted in the rice field to the bacterial blight;
a mean canopy spectral reflectance extraction module, configured to extract a mean canopy spectral reflectance of each plot in the rice field based on the multi-spectral images, wherein the rice field is divided into a plurality of plots;
a depth spectrum feature output module, configured to conduct regression of the severity of the rice bacterial blight using a convolutional neural network based on the mean canopy spectral reflectance and the severity of the rice bacterial blight for each plot, and output a depth spectrum feature;
a disease severity regression model training module, configured to train a disease severity regression model with the accumulated temperature data, the depth spectrum feature, and the resistance to the bacterial blight for each plot as an input and the corresponding severity of the rice bacterial blight as an output to obtain a well-trained disease severity regression model; and
a disease severity monitoring module, configured to input accumulated temperature data, a depth spectrum feature and resistance to the bacterial blight corresponding to each plot of a to-be-monitored rice field into the well-trained disease severity regression model to output a level of severity of the rice bacterial blight in the to-be-monitored rice field.

### Claim 8 (depends on 7)

The system according to claim 7, further comprising:
an optimal band combination determination module, configured to output an optimal band combination using an attention mechanism module in the convolutional neural network, wherein the optimal band combination is a combination of wavelengths corresponding to first N terms in a sequence obtained by sorting wavelength weights outputted by the attention mechanism module from large to small;
a rice bacterial blight monitoring SI construction module, configured to construct a monitoring SI of the rice bacterial blight according to the optimal band combination; and
a visual image determination module for the severity of the rice bacterial blight, configured to obtain a monitoring SI of the rice bacterial blight in the to-be-monitored rice field, and map the monitoring SI of the rice bacterial blight in the to-be-monitored rice field into the multi-spectral image to obtain a visual image of the severity of the rice bacterial blight in the to-be-monitored rice field.

### Claim 9 (depends on 7)

The system according to claim 7, wherein the multi-spectral image obtaining module comprises:
a remote sensing image obtaining unit, configured to obtain remote sensing images of the rice field at different growth stages by an UAV equipped with a multi-spectral camera; and
an image stitching unit, configured to stitch a plurality of the remote sensing images to obtain the multi-spectral image of the rice field as a whole.

### Claim 10 (depends on 7)

The system according to claim 7, wherein the bacterial blight severity and accumulated temperature data obtaining module comprises:
a bacterial blight severity determination unit, configured to determine the severity of the rice bacterial blight in each plot according to a proportion of an incidence area of the rice bacterial blight in a total area of each plot; and
an accumulated temperature data determination unit, configured to record maximum and minimum temperatures of each plot on a day when the multi-spectral image is obtained, and calculate the accumulated temperature data of each plot through an accumulated temperature formula.
