CN115456960A - Citrus huanglongbing disease and pest monitoring and early warning system and method - Google Patents
Citrus huanglongbing disease and pest monitoring and early warning system and method Download PDFInfo
- Publication number
- CN115456960A CN115456960A CN202211013259.0A CN202211013259A CN115456960A CN 115456960 A CN115456960 A CN 115456960A CN 202211013259 A CN202211013259 A CN 202211013259A CN 115456960 A CN115456960 A CN 115456960A
- Authority
- CN
- China
- Prior art keywords
- information
- data
- image
- module
- early warning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- Medical Informatics (AREA)
- Mathematical Optimization (AREA)
- Evolutionary Computation (AREA)
- Pure & Applied Mathematics (AREA)
- Algebra (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention provides a citrus huanglongbing disease and pest monitoring and early warning system and a method. Meanwhile, the integrated method for preventing the Huanglongbing through environmental detection and diagnosing the Huanglongbing through the near infrared spectrum combined with the image recognition technology, which is adopted by the invention, avoids the one-sided result caused by one-way diagnosis, has a wider coverage range, is suitable for large-scale planting environment, can perform nondestructive detection on plants, reduces the cost, improves the efficiency, saves the human resources, and has important social significance on the stable development of the citrus industry.
Description
Technical Field
The invention relates to the field of plant disease and insect pest monitoring and protection, in particular to a citrus huanglongbing disease and insect pest monitoring and early warning system and method.
Background
The citrus industry is the fruit planting industry with the largest cultivation area and the largest population in south China. The citrus yellow dragon disease (HLB) is one of the diseases frequently occurring in citrus planting, and has the characteristics of strong infectivity and large damage. Once infected, the citrus can spread rapidly, and no good curing method exists at present, and the citrus can only be dug out of the whole root of the infected fruit tree to avoid further spreading of insect pests. Researches show that the fruit tree diseases generally occur in the environment with excessive rainwater, large air humidity, high temperature, large soil humidity, acidic soil and insufficient sunlight irradiation. Under the environment, the citrus is easy to grow new shoots, so that the propagation and the spread of diaphorina citri are facilitated, and the diaphorina citri mainly damages tender shoots of the citrus, but carries a large amount of yellow dragon disease germs, so that the yellow dragon disease is generated. When yellow dragon disease is infected, nutrient transport of the citrus cannot be carried out due to abnormal nutrient metabolism of the citrus, absorption of mineral elements such as Ca element, mg element and Mn element is reduced, photosynthesis is weakened, and the leaves are gradually yellowed in mottle type and deficient type, so that the diseased leaves are easy to fall off, the fruits are small and deformed, the coloring is not uniform, the quality of the fruits is reduced, even the fruits are dead, and the economic benefit of citrus planting is seriously influenced.
With the growing range of citrus in recent years, the citrus greening disease has a spreading trend, which poses serious threats to the stability and development of citrus industry in China and even in the world, and the attention to the citrus greening disease needs to be increased. At present, the huanglongbing has no good treatment method, and the environmental change when the huanglongbing is easy to occur is monitored and found in time according to the pathogenesis of the huanglongbing, and the trace elements causing the huanglongbing are detected, so that the huanglongbing can be prevented better. At present, whether the huanglongbing occurs or not is judged mainly by adopting a field observation method, which is often inaccurate in diagnosis of the huanglongbing at the initial stage of the onset, and diagnosis and treatment after the onset are not beneficial to the overall prevention and treatment of the huanglongbing. The yellow dragon disease analysis in a laboratory mainly utilizes a Polymerase Chain Reaction (PCR) method, which has high detection precision but long detection time and is difficult to treat the diseased citrus in time.
The patent discloses an early warning system based on machine vision citrus greening disease, which comprises terminal monitoring nodes arranged in a citrus orchard; the coordinator is connected with the monitoring nodes and used for summarizing the collected monitoring data and sending the collected monitoring data to the core board in a serial port line connection mode; the core board is connected with the coordinator and used for processing the data information; the gateway unit is connected with the core board and is used for transmitting the processed data; the cloud server is connected with the gateway unit, and uploads the converted data information through a network after frame format conversion and protocol conversion are carried out on the monitoring data; and the database is connected with the cloud server and used for storing the data. The early warning of the citrus yellow dragon disease is realized, the red river oranges infected with the yellow dragon disease are timely and quickly discovered and identified, an important basis is provided for prediction and comprehensive diagnosis and treatment of the red river orange disease condition, and the method has very important guiding significance for reducing the loss caused by the yellow dragon disease. However, the patent does not relate to any technology which can quickly and effectively detect the pathogenic environment and influence elements of the huanglongbing, eradicates the onset of the huanglongbing from the source and has important social significance and economic value for ensuring the high-quality and healthy development of the citrus planting industry.
Disclosure of Invention
The invention provides a monitoring and early warning system for citrus huanglongbing diseases and insect pests, which can quickly and effectively detect the onset environment and influence elements of the huanglongbing.
The invention further aims to provide a monitoring method of the citrus huanglongbing disease and insect pest monitoring and early warning system.
In order to achieve the technical effects, the technical scheme of the invention is as follows:
a monitoring and early warning system for citrus huanglongbing plant diseases and insect pests comprises a plant pest monitoring and early warning module and a near infrared spectrum combined image diagnosis module; the insect pest monitoring and early warning module comprises a control terminal module, an information acquisition module, a display module, a threshold setting module and an early warning module; the near infrared spectrum combined image diagnosis module comprises an image information acquisition module, a data preprocessing module and a data analysis module;
the control terminal module is used for realizing overall detection, control of switching and monitoring and receiving of information, the information acquisition module is used for acquiring external environment information of the fruit number, the display module is used for displaying multi-source information acquired by the acquisition module in real time, and the threshold setting module is used for setting threshold information of the acquisition module; the early warning module is used for realizing early warning;
the image information acquisition module is used for acquiring the tree leaf image of the fruit tree, and the data preprocessing module is used for preprocessing the spectral data and the image information data; the data analysis module is used for analyzing the infection condition of the citrus greening disease by combining spectral diagnosis and an image recognition technology.
Furthermore, the insect pest monitoring and early warning module displays information on the display module according to the multi-source information acquired by the information acquisition module, and compares and analyzes the information with an insect pest information base stored by the terminal to obtain the type of the insect pest information suffered by the current monitoring point, so that the state of the current monitoring point is given, and the occurrence condition of the citrus yellow shoot is predicted; according to the prediction result of the disease and insect pest information monitored on site, issuing early warning information to technicians and managers, starting an early warning module, and issuing pest distribution and hazard reports; provides the safety control strategy and technical proposal of citrus greening disease, comprising pesticide spraying strategy, citrus psyllid trapping strategy and a plurality of conventional plant protection measures and techniques for eliminating diseased trees.
Furthermore, the information acquisition module adopts high-precision waterproof air temperature and humidity sensor, soil pH value sensor, leaf surface humidity sensor and light intensity sensor.
Furthermore, the display module uses an LCD display screen, and the threshold setting module is an independent key operated manually. Comparing the collected information with database information to know the change of external environment information before and after insect damage occurs, and setting corresponding threshold values according to the change to realize threshold value conversion setting and addition and subtraction operation; the early warning module uses a buzzer. When the index acquired by the information acquisition module exceeds a set threshold value, the early warning module can be automatically started to play a role in early warning; the image information acquisition module is used for acquiring the image information of the fruit trees within the early warning range after the early warning module is started, transmitting the image information to the control terminal and further analyzing and processing the image information by the terminal; the data preprocessing module is used for preprocessing the manually acquired spectral data and the image information data for better subsequent analysis when the manually acquired spectral data and the image information data are received; the data analysis module combines and unifies the manually acquired spectral data and the image information acquired data through a feature fusion technology and realizes the sickening discrimination when the manually acquired spectral data and the image information acquired data are transmitted to the terminal.
A method for monitoring and early warning citrus huanglongbing plant diseases and insect pests comprises the following steps:
s1: the control terminal module carries out threshold setting on an air temperature and humidity sensor, a soil PH value sensor, a leaf surface humidity sensor and a light intensity sensor in the information acquisition module according to the information of the insect pest database so as to timely warn the environment where insect pests occur;
s2: the information acquisition module acquires the air temperature and humidity, soil temperature and humidity, leaf surface humidity and illumination intensity of the external environment by using the sensors;
s3: the display module displays the acquired information on a screen to monitor the terminal;
s4: judging the environment where the diseases and insect pests occur according to the detection result, and if the information acquired by the information acquisition module exceeds or is lower than a set threshold value, namely the environment is favorable for the attack of the diseases and insect pests, early warning is carried out and S4 is entered; otherwise, continuing to detect;
s5: an image information acquisition module in the range of the early warning module can directly acquire the image of the fruit tree, and a worker acquires the spectrum information of trace elements of the leaves on site by using a portable near-infrared spectrometer;
s6: the data preprocessing module respectively preprocesses the spectral data and the image information;
s7: and the data analysis module analyzes the infection condition of the citrus greening disease by using a spectral diagnosis and image recognition technology.
Further, the specific process that the staff utilizes portable near-infrared spectrometer to obtain leaf microelement spectral information on the spot is:
the citrus trees in the early warning range are uniformly selected manually, leaves are randomly adopted for each tree, and spectrum collection is carried out after simple cleaning and flattening are carried out before spectrum detection operation.
Further, the process of preprocessing the spectral data is as follows:
preprocessing data acquired by a near infrared spectrum, normalizing the near infrared spectrum data and performing standard normal variable distribution processing to eliminate the influence of solid particle size and optical path change on the spectrum, correcting the spectral error of a sample due to scattering, correcting a base line, reducing the interference of temperature and moisture factors on the spectrum of a blade to a certain extent, performing Mahalanobis distance analysis on the spectral data by using MATLAB software due to the error in the spectrum acquisition process, determining a sample exceeding a Mahalanobis distance threshold as an abnormal sample, and removing an outlier in the data;
spectral data feature extraction: because the citrus leaves infected with the huanglongbing disease can change to a certain extent, the spectral characteristics of the citrus leaves are correspondingly changed and are inconsistent with the spectral data of normal leaves, and some wave bands are obviously different; the spectrum peak value of mineral elements lost from diseased leaves is obviously lower than that of normal leaves, which is because the yellow dragon disease leaves can influence the leaves to absorb nutrients and reduce the mineral element absorption capacity of the leaves; the spectral image has high correlation between adjacent bands, a large amount of redundant repeated information exists, and effective information needs to be extracted from the redundant repeated information to identify a target; performing Principal Component Analysis (PCA) technology on the preprocessed data to extract features, acquiring feature vectors, reducing dimensionality of the spectral data, reserving components with large variance and more information, discarding components with less information content, and further reducing irrelevant spectral features; the spectrum characteristic reserved after the preprocessing is called as principal component characteristic PC, the principal component represents the data structure of the original spectrum variable as much as possible without losing information, the original high-dimensional spectrum data is low-dimensional data through a characteristic extraction technology, and meanwhile, certain necessary characteristics of the original data are reserved, so that dimension explosion can be avoided to a great extent, tasks such as subsequent classification or clustering are more stable and easy to process, and more importantly, more excellent generalization performance is generated.
Further, the specific process of preprocessing the image information is as follows:
preprocessing a citrus leaf image acquired by a camera, wherein the image is influenced by different noise sources in the acquisition process, including random noise generated in an image acquisition device and an image transmission process, and is subjected to denoising by utilizing wavelet transformation and a median filtering technology, so that the quality of the leaf image is optimized, and the feature extraction is further performed better;
image data feature extraction:
carrying out gray processing on the leaf image by adopting a traditional image processing method, and deepening the difference between a target and a background by improving the component of a gray image according to the obvious difference between a leaf and the background; and carry out binarization processing when calculating RGB component, improve computational efficiency. Namely, by the traditional image processing method, a color image of three channels is input, and a gray scale image is output; the method comprises the following steps of (1) adopting a Sobal algorithm to carry out edge segmentation on a blade, wherein the gradient of a pixel value is calculated by utilizing the change characteristics of target edge pixels and background pixels due to the fuzzy background, clear target blade pixel points and obvious edge gray difference, and the position with large gradient change amplitude is the edge of an image; boundary segmentation is carried out by utilizing boundary information, a blade background is eliminated, and the blade outline and the characteristics of the blade are kept and taken as the shape characteristics of the image; the color information of the leaves is extracted by using a color extraction algorithm, and any one color can be obtained by adding and mixing three colors of three primary colors of red, green and blue according to different components.
Further, extracting the average of the three primary colors and the standard deviation thereof, using the average as one of color characteristics of the leaf, converting the RGB color domain into the HSV color domain, and using the color domain mean and the standard deviation of the HSV color as the color characteristics; the texture features do not depend on color or brightness, and reflect local irregularity on the image and macroscopic regular characteristics; processing the RGB image into a gray image by adopting a gray difference method, analyzing and extracting 4 texture characteristic parameters by adopting an image gray co-occurrence matrix method, respectively extracting four characteristic parameters of energy, entropy, moment of inertia and correlation, and taking the average and standard deviation as texture characteristics; and carrying out normalization processing on the shape characteristic variable, the color characteristic variable and the texture characteristic variable of the image by adopting a normalization method.
Further, the principal component analysis technique comprises the steps of:
principal component analysis method is characterized by using many original variables X with correlation 1 ,X 2 ,...,X P Reestablishing a group of less and mutually irrelevant comprehensive indexes F m Instead of the original variables:
2) And data preprocessing:
in order to eliminate the difference of various data characteristics on the magnitude, the data are standardized to obtain a standardized matrix;
2) Calculating covariance matrix sigma =(s) ij ) P×P ：
And (3) establishing a covariance matrix according to the standardized data matrix, wherein the covariance matrix is an evaluation index reflecting the correlation degree between the standardized data, and the larger the value is, the more n is the number of data samples, which indicates that the main component analysis of the data is necessary:
3) Calculating the eigenvalue and the eigenvector of the covariance matrix
Carrying out orthogonal decomposition on the covariance matrix according to the established covariance matrix, and calculating an eigenvalue and an eigenvector to obtain a principal component; the first m larger eigenvalues λ in Σ 1 >λ 2 >…>λ m >0, i.e. the variance, λ, corresponding to the first m principal components i Eigenvector alpha corresponding to eigenvalue i Is a main component F i With respect to the correlation coefficient of the original variable, the ith principal component F of the original variable i Comprises the following steps: f i ＝α i ·X
The variance contribution rate of the principal component reflects the magnitude of its information content α i Comprises the following steps:
4) Selecting the main component
Determining the number of principal components, i.e. F 1 ，F 2 .....F m The determination of m is determined by the variance cumulative contribution rate G (m), and when the cumulative contribution rate is greater than 85%, it is considered that the information of the original variable is reflected:
5) Calculating principal component load
Principal component loading is a reflection of principal component F i With the original variable X j Degree of correlation between them, original variable X j In the main component F i Load on L ij ：
6) Calculating the principal component value
Calculating the scores of the data on the m main components to obtain a comprehensive index F i ：
F i ＝α 1i X 1 +α 2i X 2 +…+α Pi X P 。
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
according to the citrus huanglongbing monitoring system, the problems of the prior art and product modularization are solved through the insect pest monitoring and early warning module and the near infrared spectrum combined image diagnosis module. Meanwhile, the integrated method for preventing the huanglongbing through environmental detection and diagnosing the huanglongbing through the near infrared spectrum combined with the image recognition technology, which is adopted by the invention, avoids the too one-sided result caused by one-way diagnosis, has wider coverage range, is suitable for large-scale planting environment, can perform nondestructive detection on plants, reduces the cost, improves the efficiency, saves the human resources, and has important social significance for the stable development of the citrus industry.
Drawings
FIG. 1 is a block diagram of the system of the present invention;
FIG. 2 is a flow chart of the method of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described with reference to the drawings and the embodiments.
Example 1
As shown in fig. 1, a monitoring and early warning system for citrus huanglongbing plant diseases and insect pests comprises a pest monitoring and early warning module and a near infrared spectrum combined image diagnosis module; the insect pest monitoring and early warning module comprises a control terminal module, an information acquisition module, a display module, a threshold setting module and an early warning module; the near infrared spectrum combined image diagnosis module comprises an image information acquisition module, a data preprocessing module and a data analysis module;
the control terminal module is used for realizing overall detection, control of switching and monitoring and receiving of information, the information acquisition module is used for acquiring external environment information of the fruit number, the display module is used for displaying multi-source information acquired by the acquisition module in real time, and the threshold setting module is used for setting threshold information of the acquisition module; the early warning module is used for realizing early warning;
the image information acquisition module is used for acquiring the tree leaf image of the fruit tree, and the data preprocessing module is used for preprocessing the spectral data and the image information data; the data analysis module is used for analyzing the infection condition of the citrus greening disease by combining spectral diagnosis with an image recognition technology.
The pest monitoring and early warning module displays information on the display module according to the multi-source information acquired by the information acquisition module, and compares and analyzes the information with a pest information base stored by the terminal to obtain the type of pest information suffered by a current monitoring point, so that the state of the current monitoring point is given, and the occurrence condition of citrus yellow shoot is predicted; according to the prediction result of the disease and insect pest information monitored on site, issuing early warning information to technicians and managers, starting an early warning module, and issuing pest distribution and hazard reports; provides safety control strategies and technical suggestions of citrus greening disease, including pesticide spraying strategies, citrus psyllid trapping strategies and a plurality of conventional plant protection measures and technologies for removing diseased trees.
The information acquisition module adopts high accuracy waterproof air temperature and humidity sensor, soil pH value sensor, leaf surface humidity sensor and light intensity sensor.
The display module uses an LCD display screen, and the threshold setting module is an independent key operated manually. Comparing the collected information with database information to know the change of external environment information before and after insect damage occurs, and setting corresponding threshold values according to the change to realize threshold value conversion setting and addition and subtraction operation; the early warning module uses a buzzer. When the index acquired by the information acquisition module exceeds a set threshold value, the early warning module can be automatically started to play a role in early warning; the image information acquisition module is used for acquiring the image information of the fruit trees within the early warning range after the early warning module is started, transmitting the image information to the control terminal and further analyzing and processing the image information by the terminal; the data preprocessing module is used for preprocessing the manually acquired spectral data and the image information data for better subsequent analysis when the manually acquired spectral data and the image information data are received; the data analysis module is used for merging and unifying the manually acquired spectral data and the data acquired by the image information by a characteristic fusion technology to realize the sickening judgment when the manually acquired spectral data and the data acquired by the image information are transmitted to the terminal.
Example 2
As shown in fig. 2, the monitoring and early warning method for citrus huanglongbing plant diseases and insect pests comprises the following steps:
s1: the control terminal module carries out threshold setting on an air temperature and humidity sensor, a soil PH value sensor and a leaf surface humidity sensor in the information acquisition module according to the information of the insect pest database so as to early warn the environment where insect pests occur in time;
s2: the information acquisition module acquires the air temperature and humidity, the soil temperature and humidity, the leaf surface humidity and the illumination intensity of the external environment by using the sensors;
s3: the display module displays the acquired information on a screen to monitor the terminal;
s4: judging the environment where the diseases and insect pests occur according to the detection result, if the information acquired by the information acquisition module exceeds or is lower than a set threshold value, namely the environment is favorable for the attack of the diseases and insect pests, early warning is carried out, and S4 is entered; otherwise, continuing to detect;
s5: an image information acquisition module in the range of the early warning module can directly acquire the image of the fruit tree, and a worker acquires the spectrum information of trace elements of the leaves on site by using a portable near-infrared spectrometer;
s6: the data preprocessing module respectively preprocesses the spectral data and the image information;
s7: and the data analysis module analyzes the infection condition of the citrus greening disease by using a spectrum diagnosis and an image recognition technology.
Further, the specific process that the staff utilizes portable near-infrared spectrometer to obtain leaf microelement spectral information on the spot is:
the citrus trees in the early warning range are uniformly selected manually, leaves are randomly adopted for each tree, and spectrum collection is carried out after simple cleaning and flattening are carried out before spectrum detection operation.
The process of preprocessing the spectral data is as follows:
preprocessing data acquired by a near infrared spectrum, normalizing the near infrared spectrum data and performing standard normal variable distribution processing to eliminate the influence of solid particle size and optical path change on the spectrum, correcting the spectral error of a sample due to scattering, correcting a base line, reducing the interference of temperature and moisture factors on the spectrum of a blade to a certain extent, performing Mahalanobis distance analysis on the spectral data by using MATLAB software due to the error in the spectrum acquisition process, determining a sample exceeding a Mahalanobis distance threshold as an abnormal sample, and removing an outlier in the data;
spectral data feature extraction: because the citrus leaves infected with the huanglongbing disease can change to a certain extent, the spectral characteristics of the citrus leaves are correspondingly changed and are inconsistent with the spectral data of normal leaves, and some wave bands are obviously different; the spectral peak value of mineral elements missing from diseased leaves is obviously lower than that of normal leaves, because the yellow dragon diseased leaves can influence the leaves to absorb nutrients, and the capacity of the leaves to absorb the mineral elements is reduced; the spectral image has high correlation between adjacent bands, a large amount of redundant repeated information exists, and effective information needs to be extracted from the redundant repeated information to identify a target; performing Principal Component Analysis (PCA) technology on the preprocessed data to extract features, acquiring feature vectors, reducing dimensionality of the spectral data, reserving components with large variance and more information, discarding components with less information content, and further reducing irrelevant spectral features; the spectrum characteristic reserved after the preprocessing is called as principal component characteristic PC, the principal component represents the data structure of the original spectrum variable as much as possible without losing information, the original high-dimensional spectrum data is low-dimensional data through a characteristic extraction technology, and meanwhile, certain necessary characteristics of the original data are reserved, so that dimension explosion can be avoided to a great extent, tasks such as subsequent classification or clustering are more stable and easy to process, and more importantly, more excellent generalization performance is generated.
The specific process of preprocessing the image information is as follows:
preprocessing a citrus leaf image acquired by a camera, wherein the image is influenced by different noise sources in the acquisition process, including random noise generated in an image acquisition device and an image transmission process, and is subjected to denoising by utilizing wavelet transformation and a median filtering technology, so that the quality of the leaf image is optimized, and the feature extraction is further performed better;
image data feature extraction:
carrying out gray processing on the leaf image by adopting a traditional image processing method, and deepening the difference between a target and a background by improving the component of a gray image according to the obvious difference between a leaf and the background; and carry out binarization processing when calculating RGB component, improve computational efficiency. Namely, by the traditional image processing method, a color image of three channels is input, and a gray scale image is output; the method comprises the following steps of (1) adopting a Sobal algorithm to carry out edge segmentation on a blade, wherein the gradient of a pixel value is calculated by utilizing the change characteristics of target edge pixels and background pixels due to the fuzzy background, clear target blade pixel points and obvious edge gray difference, and the position with large gradient change amplitude is the edge of an image; boundary segmentation is carried out by utilizing boundary information, a leaf background is eliminated, and leaf outlines and self characteristics are reserved and are used as shape characteristics of the image; the color information of the leaves is extracted by using a color extraction algorithm, and any one color can be obtained by adding and mixing three colors of three primary colors of red, green and blue according to different components.
Extracting the average of three primary colors and the standard deviation thereof, taking the average as one of color characteristics of the leaf, converting an RGB color domain into an HSV color domain, and taking the color domain average and the standard deviation of the HSV color as the color characteristics; the texture features do not depend on color or brightness, and reflect local irregularity on the image and macroscopic regular characteristics; processing the RGB image into a gray image by adopting a gray difference method, analyzing and extracting 4 texture characteristic parameters by adopting an image gray co-occurrence matrix method, respectively extracting four characteristic parameters of energy, entropy, moment of inertia and correlation, and taking the average and standard deviation as texture characteristics; and carrying out normalization processing on the shape characteristic variable, the color characteristic variable and the texture characteristic variable of the image by adopting a normalization method.
The principal component analysis technique comprises the following steps:
the idea of principal component analysis is to use a plurality of original variables X with correlation 1 ,X 2 ,...,X P Reestablishing a group of less and mutually irrelevant comprehensive indexes F m Instead of the original variables:
3) And data preprocessing:
in order to eliminate the difference of various data characteristics on the magnitude, the data are standardized to obtain a standardized matrix;
2) Calculating covariance matrix sigma =(s) ij ) P×P ：
And establishing a covariance matrix according to the standardized data matrix, wherein the covariance matrix is an evaluation index reflecting the correlation degree between the standardized data, and the larger the value is, the more n is the number of data samples, which indicates that the main component analysis of the data is necessary:
3) Calculating the eigenvalue and the eigenvector of the covariance matrix
Carrying out orthogonal decomposition on the covariance matrix according to the established covariance matrix, and calculating an eigenvalue and an eigenvector to obtain a principal component; the first m larger eigenvalues λ in Σ 1 >λ 2 >…>λ m >0, i.e. the variance, λ, corresponding to the first m principal components i Eigenvector alpha corresponding to eigenvalue i Is a main component F i With respect to the correlation coefficient of the original variable, the ith principal component F of the original variable i Comprises the following steps: f i ＝α i ·X
The variance contribution rate of the principal component reflects the magnitude of its information amount α i Comprises the following steps:
4) Selecting the main component
Determining the number of principal components, i.e. F 1 ，F 2 .....F m The determination of m is determined by the variance cumulative contribution rate G (m), and when the cumulative contribution rate is greater than 85%, it is considered that the information of the original variable is reflected:
5) Calculating principal component load
Principal componentThe partial load being a reflection of the principal component F i With the original variable X j Degree of correlation between them, original variable X j In the main component F i Load on L ij ：
6) Calculating the principal component value
Calculating the scores of the data on the m main components to obtain a comprehensive index F i ：
F i ＝α 1i X 1 +α 2i X 2 +…+α Pi X P 。
Example 3
As shown in fig. 1, a monitoring and early warning system for citrus huanglongbing plant diseases and insect pests comprises a pest monitoring and early warning module and a near infrared spectrum combined image diagnosis module; the insect pest monitoring and early warning module comprises a control terminal module, an information acquisition module, a display module, a threshold setting module and an early warning module; the near infrared spectrum combined image diagnosis module comprises an image information acquisition module, a data preprocessing module and a data analysis module;
the control terminal module is used for realizing overall detection, control of switching and monitoring and receiving of information, the information acquisition module is used for acquiring external environment information of the fruit number, the display module is used for displaying multi-source information acquired by the acquisition module in real time, and the threshold setting module is used for setting threshold information of the acquisition module; the early warning module is used for realizing early warning;
the image information acquisition module is used for acquiring the tree leaf image of the fruit tree, and the data preprocessing module is used for preprocessing the spectral data and the image information data; the data analysis module is used for analyzing the infection condition of the citrus greening disease by combining spectral diagnosis with an image recognition technology.
The pest monitoring and early warning module displays information on the display module according to the multi-source information acquired by the information acquisition module, and compares and analyzes the information with a pest information base stored by the terminal to obtain the type of pest information suffered by a current monitoring point, so that the state of the current monitoring point is given, and the occurrence condition of citrus yellow shoot is predicted; according to the prediction result of the disease and insect pest information monitored on site, issuing early warning information to technicians and managers, starting an early warning module, and issuing pest distribution and hazard reports; provides safety control strategies and technical suggestions of citrus greening disease, including pesticide spraying strategies, citrus psyllid trapping strategies and a plurality of conventional plant protection measures and technologies for removing diseased trees.
The information acquisition module adopts high-precision waterproof air temperature and humidity sensor, soil pH value sensor, leaf surface humidity sensor and light intensity sensor.
The display module uses an LCD display screen, and the threshold setting module is an independent key operated manually. Comparing the collected information with database information to know the change of external environment information before and after insect damage occurs, and setting corresponding threshold values according to the change to realize threshold value conversion setting and addition and subtraction operation; the early warning module uses a buzzer. When the index acquired by the information acquisition module exceeds a set threshold value, the early warning module can be automatically started to play a role in early warning; the image information acquisition module is used for acquiring the image information of the fruit trees within the early warning range after the early warning module is started, transmitting the image information to the control terminal and further analyzing and processing the image information by the terminal; the data preprocessing module is used for preprocessing the manually acquired spectral data and the image information data for better subsequent analysis when the manually acquired spectral data and the image information data are received; the data analysis module is used for merging and unifying the manually acquired spectral data and the data acquired by the image information by a characteristic fusion technology to realize the sickening judgment when the manually acquired spectral data and the data acquired by the image information are transmitted to the terminal.
As shown in fig. 2, the monitoring and early warning method of the monitoring and early warning system for citrus huanglongbing plant diseases and insect pests includes the following steps:
s1: the control terminal module carries out threshold setting on an air temperature and humidity sensor, a soil PH value sensor and a leaf surface humidity sensor in the information acquisition module according to the information of the insect pest database so as to early warn the environment where insect pests occur in time;
s2: the information acquisition module acquires the air temperature and humidity, the soil temperature and humidity, the leaf surface humidity and the illumination intensity of the external environment by using the sensors;
s3: the display module displays the acquired information on a screen for monitoring the terminal;
s4: judging the environment where the diseases and insect pests occur according to the detection result, if the information acquired by the information acquisition module exceeds or is lower than a set threshold value, namely the environment is favorable for the attack of the diseases and insect pests, early warning is carried out, and S4 is entered; otherwise, continuing to detect;
s5: an image information acquisition module in the range of the early warning module can directly acquire the image of the fruit tree, and a worker acquires the spectrum information of trace elements of the leaves on site by using a portable near-infrared spectrometer;
s6: the data preprocessing module respectively preprocesses the spectral data and the image information;
s7: and the data analysis module analyzes the infection condition of the citrus greening disease by using a spectrum diagnosis and an image recognition technology.
Further, the specific process that the staff utilizes portable near-infrared spectrometer to obtain leaf microelement spectral information on the spot is:
the citrus trees in the early warning range are uniformly selected manually, leaves are randomly adopted for each tree, and spectrum collection is carried out after simple cleaning and flattening are carried out before spectrum detection operation.
The process of preprocessing the spectral data is as follows:
preprocessing data acquired by a near infrared spectrum, normalizing the near infrared spectrum data and performing standard normal variable distribution processing to eliminate the influence of solid particle size and optical path change on the spectrum, correcting the spectral error of a sample due to scattering, correcting a base line, reducing the interference of temperature and moisture factors on the spectrum of a blade to a certain extent, performing Mahalanobis distance analysis on the spectral data by using MATLAB software due to the error in the spectrum acquisition process, determining a sample exceeding a Mahalanobis distance threshold as an abnormal sample, and removing an outlier in the data;
spectral data feature extraction: because the citrus leaves infected with the huanglongbing disease can change to a certain extent, the spectral characteristics of the citrus leaves are correspondingly changed and are inconsistent with the spectral data of normal leaves, and some wave bands are obviously different; the spectral peak value of mineral elements missing from diseased leaves is obviously lower than that of normal leaves, because the yellow dragon diseased leaves can influence the leaves to absorb nutrients, and the capacity of the leaves to absorb the mineral elements is reduced; the spectral image has high correlation between adjacent bands, a large amount of redundant repeated information exists, and effective information needs to be extracted from the redundant repeated information to identify a target; performing Principal Component Analysis (PCA) technology on the preprocessed data to extract features, acquiring feature vectors, reducing dimensionality of the spectral data, reserving components with large variance and more information, discarding components with less information content, and further reducing irrelevant spectral features; the spectrum characteristics remained after the pretreatment are called as principal component characteristics PC, the principal component represents the data structure of the original spectrum variable as much as possible without losing information, the original high-dimensional spectrum data is the low-dimensional data through a characteristic extraction technology, and meanwhile, certain necessary characteristics of the original data are still remained, so that dimension explosion can be avoided to a great extent, and tasks such as subsequent classification or clustering are more stable and easy to process, and more importantly, more excellent generalization performance is generated.
The specific process of preprocessing the image information is as follows:
preprocessing a citrus leaf image acquired by a camera, wherein the image is influenced by different noise sources in the acquisition process, including random noise generated in an image acquisition device and an image transmission process, and is subjected to denoising by utilizing wavelet transformation and a median filtering technology, so that the quality of the leaf image is optimized, and the feature extraction is further performed better;
image data feature extraction:
carrying out gray processing on the leaf image by adopting a traditional image processing method, and deepening the difference between a target and a background by improving the component of a gray image according to the obvious difference between a leaf and the background; and carry out binarization processing when calculating RGB component, improve computational efficiency. Namely, by the traditional image processing method, a color image of three channels is input, and a gray scale image is output; the method comprises the following steps of (1) adopting a Sobal algorithm to carry out edge segmentation on a blade, wherein the gradient of a pixel value is calculated by utilizing the change characteristics of target edge pixels and background pixels due to the fuzzy background, clear target blade pixel points and obvious edge gray difference, and the position with large gradient change amplitude is the edge of an image; boundary segmentation is carried out by utilizing boundary information, a blade background is eliminated, and the blade outline and the characteristics of the blade are kept and taken as the shape characteristics of the image; the color information of the leaves is extracted by using a color extraction algorithm, and any one color can be obtained by adding and mixing three colors of three primary colors of red, green and blue according to different components.
Extracting the average of three primary colors and the standard deviation thereof, taking the average as one of color characteristics of the leaf, converting an RGB color domain into an HSV color domain, and taking the color domain average and the standard deviation of the HSV color as the color characteristics; the texture features do not depend on color or brightness, and reflect local irregularity on the image and macroscopic regular characteristics; processing the RGB image into a gray image by adopting a gray difference method, analyzing and extracting 4 texture characteristic parameters by adopting an image gray co-occurrence matrix method, respectively extracting four characteristic parameters of energy, entropy, moment of inertia and correlation, and taking the average and standard deviation as texture characteristics; and carrying out normalization processing on the shape characteristic variable, the color characteristic variable and the texture characteristic variable of the image by adopting a normalization method.
The principal component analysis technique comprises the following steps:
the idea of principal component analysis is to use a plurality of original variables X with correlation 1 ,X 2 ,...,X P Reestablishing a group of less and mutually irrelevant comprehensive indexes F m Instead of the original variables:
4) And data preprocessing:
in order to eliminate the difference of various data characteristics on the magnitude, the data are standardized to obtain a standardized matrix;
2) Calculating covariance matrix sigma =(s) ij ) P×P ：
And (3) establishing a covariance matrix according to the standardized data matrix, wherein the covariance matrix is an evaluation index reflecting the correlation degree between the standardized data, and the larger the value is, the more n is the number of data samples, which indicates that the main component analysis of the data is necessary:
3) Calculating the eigenvalue and the eigenvector of the covariance matrix
Carrying out orthogonal decomposition on the covariance matrix according to the established covariance matrix, and calculating an eigenvalue and an eigenvector to obtain a principal component; the first m larger eigenvalues λ in Σ 1 >λ 2 >…>λ m >0, i.e. the variance, λ, corresponding to the first m principal components i Eigenvector alpha corresponding to eigenvalue i Is a main component F i With respect to the correlation coefficient of the original variable, the ith principal component F of the original variable i Comprises the following steps: f i ＝α i ·X
The variance contribution rate of the principal component reflects the magnitude of its information content α i Comprises the following steps:
4) Selecting the main component
Determining the number of principal components, i.e. F 1 ，F 2 .....F m The determination of m is determined by the variance cumulative contribution rate G (m), and when the cumulative contribution rate is greater than 85%, it is considered that the information of the original variable is reflected:
5) Calculating principal component load
Principal component loading is a reflection of principal component F i With the original variable X j Degree of correlation between them, original variable X j In the main component F i Load on L ij ：
6) Calculating the principal component value
Calculating the scores of the data on the m main components to obtain a comprehensive index F i ：
F i ＝α 1i X 1 +α 2i X 2 +…+α Pi X P 。
The same or similar reference numerals correspond to the same or similar parts;
the positional relationships depicted in the drawings are for illustrative purposes only and are not to be construed as limiting the present patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. This need not be, nor should it be exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (10)
1. A monitoring and early warning system for citrus huanglongbing plant diseases and insect pests is characterized by comprising a plant pest monitoring and early warning module and a near infrared spectrum combined image diagnosis module; the insect pest monitoring and early warning module comprises a control terminal module, an information acquisition module, a display module, a threshold setting module and an early warning module; the near infrared spectrum combined image diagnosis module comprises an image information acquisition module, a data preprocessing module and a data analysis module;
the control terminal module is used for realizing integral detection, control of switching and monitoring and receiving of information, the information acquisition module is used for acquiring external environment information of the fruit number, the display module is used for displaying multi-source information acquired by the acquisition module in real time, and the threshold setting module is used for setting threshold information of the acquisition module; the early warning module is used for realizing early warning;
the image information acquisition module is used for acquiring the leaf images of the fruit trees, and the data preprocessing module is used for preprocessing the spectral data and the image information data; the data analysis module is used for analyzing the infection condition of the citrus greening disease by combining spectral diagnosis with an image recognition technology.
2. The citrus Huanglong pest monitoring and early warning system according to claim 1, characterized in that the pest monitoring and early warning module displays information on the display module according to the multi-source information acquired by the information acquisition module, and compares the information with a pest information base stored by the terminal for analysis to obtain the type of pest information suffered by a current monitoring point, so that the state of the current monitoring point is given, and the occurrence of citrus Huanglong disease is predicted; according to the prediction result of the pest and disease information monitored on site, issuing early warning information to technicians and managers, starting an early warning module, and issuing pest distribution and hazard reports; provides safety control strategies and technical suggestions of citrus greening disease, including pesticide spraying strategies, citrus psyllid trapping strategies and a plurality of conventional plant protection measures and technologies for removing diseased trees.
3. The monitoring and early warning system for citrus huanglongbing pests and diseases according to claim 2, characterized in that the information acquisition module adopts a high-precision waterproof air temperature and humidity sensor, a soil pH value sensor, a leaf surface humidity sensor and a light intensity sensor.
4. The citrus huanglongbing disease and pest monitoring and early warning system according to claim 3, wherein the display module uses an LCD display screen, the threshold setting module is an independent key operated manually, changes of external environment information before and after insect pests occur are known by comparing acquired information with database information, and corresponding thresholds are set through the changes to realize threshold conversion setting and addition and subtraction operation; the early warning module uses a buzzer; when the index acquired by the information acquisition module exceeds a set threshold value, the early warning module can be automatically started to play a role in early warning; the image information acquisition module is used for acquiring the image information of the fruit trees within the early warning range after the early warning module is started, transmitting the image information to the control terminal and further analyzing and processing the image information by the terminal; the data preprocessing module is used for preprocessing the manually acquired spectral data and the image information data for better subsequent analysis when the manually acquired spectral data and the image information data are received; the data analysis module is used for merging and unifying the manually acquired spectral data and the data acquired by the image information by a characteristic fusion technology to realize the sickening judgment when the manually acquired spectral data and the data acquired by the image information are transmitted to the terminal.
5. A method for monitoring and early warning of citrus huanglongbing diseases and insect pests is characterized by comprising the following steps:
s1: the control terminal module carries out threshold setting on an air temperature and humidity sensor, a soil PH value sensor and a leaf surface humidity sensor in the information acquisition module according to the information of the insect pest database so as to early warn the environment where insect pests occur in time;
s2: the information acquisition module acquires the air temperature and humidity, soil temperature and humidity, leaf surface humidity and illumination intensity of the external environment by using the sensors;
s3: the display module displays the acquired information on a screen to monitor the terminal;
s4: judging the environment where the diseases and insect pests occur according to the detection result, and if the information acquired by the information acquisition module exceeds or is lower than a set threshold value, namely the environment is favorable for the attack of the diseases and insect pests, early warning is carried out and S4 is entered; otherwise, continuing to detect;
s5: an image information acquisition module in the range of the early warning module can directly acquire the image of the fruit tree, and a worker acquires the spectrum information of trace elements of leaves on site by using a portable near-infrared spectrometer;
s6: the data preprocessing module respectively preprocesses the spectral data and the image information;
s7: and the data analysis module analyzes the infection condition of the citrus greening disease by using a spectrum diagnosis and an image recognition technology.
6. The citrus Huanglong pest and disease monitoring and early warning method according to claim 5, is characterized in that the specific process of acquiring the trace element spectral information of leaves by a worker on site by using a portable near-infrared spectrometer is as follows:
the citrus trees in the early warning range are uniformly selected manually, leaves are randomly adopted for each tree, and spectrum collection is carried out after simple cleaning and flattening are carried out before spectrum detection operation.
7. The method for monitoring and early warning of citrus huanglongbing pests and diseases according to claim 6, wherein the pretreatment process of the spectral data comprises the following steps:
preprocessing data acquired by a near infrared spectrum, normalizing the near infrared spectrum data and performing standard normal variable distribution processing to eliminate the influence of solid particle size and optical path change on the spectrum, correcting the spectral error of a sample due to scattering, correcting a base line, reducing the interference of temperature and moisture factors on the spectrum of a blade to a certain extent, performing Mahalanobis distance analysis on the spectral data by using MATLAB software due to the error in the spectrum acquisition process, determining a sample exceeding a Mahalanobis distance threshold as an abnormal sample, and removing an outlier in the data;
spectral data feature extraction: because the citrus leaves infected with the huanglongbing disease can change to a certain extent, the spectral characteristics of the citrus leaves are correspondingly changed and are inconsistent with the spectral data of normal leaves, and some wave bands are obviously different; the spectral peak value of mineral elements missing from diseased leaves is obviously lower than that of normal leaves, because the yellow dragon diseased leaves can influence the leaves to absorb nutrients, and the capacity of the leaves to absorb the mineral elements is reduced; the spectral image has high correlation between adjacent bands, a large amount of redundant repeated information exists, and effective information needs to be extracted from the redundant repeated information to identify a target; performing Principal Component Analysis (PCA) technology on the preprocessed data to extract features, acquiring feature vectors, reducing dimensionality of spectral data, reserving components with large variance and much information, discarding components with small information content, and further reducing irrelevant spectral features; the spectrum characteristic reserved after the preprocessing is called as principal component characteristic PC, the principal component represents the data structure of the original spectrum variable as much as possible without losing information, the original high-dimensional spectrum data is low-dimensional data through a characteristic extraction technology, and meanwhile, certain necessary characteristics of the original data are reserved, so that dimension explosion can be avoided to a great extent, tasks such as subsequent classification or clustering are more stable and easy to process, and more importantly, more excellent generalization performance is generated.
8. The citrus huanglongbing pest and disease monitoring and early warning method according to claim 7, characterized in that the specific process of preprocessing the image information is as follows:
preprocessing a citrus leaf image acquired by a camera, wherein the image is influenced by different noise sources in the acquisition process, including random noise generated in an image acquisition device and an image transmission process, and is subjected to denoising by utilizing wavelet transformation and a median filtering technology, so that the quality of the leaf image is optimized, and the feature extraction is further performed better;
image data feature extraction:
carrying out gray processing on the leaf image by adopting a traditional image processing method, and deepening the difference between a target and a background by improving the component of a gray image according to the obvious difference between a leaf and the background; and carry on the binarization processing while calculating RGB component, raise the computational efficiency, namely through the traditional image processing method, input is the colored drawing of the three channels, output is the gray map; the method comprises the following steps of (1) adopting a Sobal algorithm to carry out edge segmentation on a blade, wherein the gradient of a pixel value is calculated by utilizing the change characteristics of target edge pixels and background pixels due to the fuzzy background, clear target blade pixel points and obvious edge gray difference, and the position with large gradient change amplitude is the edge of an image; boundary segmentation is carried out by utilizing boundary information, a blade background is eliminated, and the blade outline and the characteristics of the blade are kept and taken as the shape characteristics of the image; the color information of the leaves is extracted by using a color extraction algorithm, and any one color can be obtained by adding and mixing three colors of three primary colors of red, green and blue according to different components.
9. The citrus huanglongbing disease and pest monitoring and early warning method according to claim 8, characterized in that the average number and standard deviation of three primary colors are extracted as one of color features of leaves, the RGB color gamut is converted into an HSV color gamut, and the color gamut mean value and standard deviation of HSV are used as color features; the texture features do not depend on color or brightness, and reflect local irregularity on the image and macroscopic regular characteristics; processing the RGB image into a gray image by adopting a gray difference method, analyzing and extracting 4 texture characteristic parameters by adopting an image gray co-occurrence matrix method, respectively extracting four characteristic parameters of energy, entropy, moment of inertia and correlation, and taking the average and standard deviation as texture characteristics; and carrying out normalization processing on the shape characteristic variable, the color characteristic variable and the texture characteristic variable of the image by adopting a normalization method.
10. The method for monitoring and warning citrus huanglongbing pests and diseases according to claim 9, wherein the Principal Component Analysis (PCA) technology comprises the following steps:
the idea of principal component analysis is to use a plurality of original variables X with correlation 1 ,X 2 ,...,X P Reestablishing a group of less and mutually irrelevant comprehensive indexes F m Instead of the original variables:
1) And data preprocessing:
in order to eliminate the difference of various data characteristics on the magnitude, the data are standardized to obtain a standardized matrix;
2) Calculating covariance matrix sigma =(s) ij ) P×P ：
And (3) establishing a covariance matrix according to the standardized data matrix, wherein the covariance matrix is an evaluation index reflecting the correlation degree between the standardized data, and the larger the value is, the more n is the number of data samples, which indicates that the main component analysis of the data is necessary:
3) Calculating the eigenvalue and the eigenvector of the covariance matrix
Carrying out orthogonal decomposition on the covariance matrix according to the established covariance matrix, and calculating an eigenvalue and an eigenvector to obtain a principal component; the first m larger eigenvalues λ in Σ 1 >λ 2 >…>λ m >0, i.e. the variance, λ, corresponding to the first m principal components i Eigenvector alpha corresponding to eigenvalue i Is a main component F i With respect to the correlation coefficient of the original variable, the ith principal component F of the original variable i Comprises the following steps: f i ＝α i ·X
The variance contribution rate of the principal component reflects the magnitude of its information content α i Comprises the following steps:
4) Selecting the main component
Determining the number of principal components, i.e. F 1 ，F 2 .....F m The determination of m is determined by the variance cumulative contribution rate G (m), and when the cumulative contribution rate is greater than 85%, it is considered that the information of the original variable is reflected:
5) Calculating principal component load
Principal component loading is a reflection of principal component F i With the original variable X j Degree of correlation between them, original variable X j In the main component F i Load on L ij ：
6) Calculating the principal component value
Calculating the scores of the data on the m main components to obtain a comprehensive index F i ：
F i ＝α 1i X 1 +α 2i X 2 +…+α Pi X P 。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211013259.0A CN115456960A (en) | 2022-08-23 | 2022-08-23 | Citrus huanglongbing disease and pest monitoring and early warning system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211013259.0A CN115456960A (en) | 2022-08-23 | 2022-08-23 | Citrus huanglongbing disease and pest monitoring and early warning system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115456960A true CN115456960A (en) | 2022-12-09 |
Family
ID=84297905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211013259.0A Pending CN115456960A (en) | 2022-08-23 | 2022-08-23 | Citrus huanglongbing disease and pest monitoring and early warning system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115456960A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115953691A (en) * | 2023-03-09 | 2023-04-11 | 昆明理工大学 | Method and system for dynamically identifying earthwork blasting |
| CN116258373A (en) * | 2023-03-15 | 2023-06-13 | 杭州盈禾嘉田科技有限公司 | Disease and pest detection, prediction and early warning system and method based on big data |
| CN117557914A (en) * | 2024-01-08 | 2024-02-13 | 成都大学 | Crop pest identification method based on deep learning |
| CN117807548A (en) * | 2024-02-29 | 2024-04-02 | 江苏新希望生态科技有限公司 | Bean sprout growth and cultivation environment monitoring method |
| CN118037716A (en) * | 2024-04-11 | 2024-05-14 | 贵州省园艺研究所(贵州省园艺工程技术研究中心) | Cowpea pest and disease damage identification method and system for field scene |
-
2022
- 2022-08-23 CN CN202211013259.0A patent/CN115456960A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115953691A (en) * | 2023-03-09 | 2023-04-11 | 昆明理工大学 | Method and system for dynamically identifying earthwork blasting |
| CN116258373A (en) * | 2023-03-15 | 2023-06-13 | 杭州盈禾嘉田科技有限公司 | Disease and pest detection, prediction and early warning system and method based on big data |
| CN116258373B (en) * | 2023-03-15 | 2024-02-09 | 杭州盈禾嘉田科技有限公司 | Disease and pest detection, prediction and early warning system and method based on big data |
| CN117557914A (en) * | 2024-01-08 | 2024-02-13 | 成都大学 | Crop pest identification method based on deep learning |
| CN117557914B (en) * | 2024-01-08 | 2024-04-02 | 成都大学 | Crop pest identification method based on deep learning |
| CN117807548A (en) * | 2024-02-29 | 2024-04-02 | 江苏新希望生态科技有限公司 | Bean sprout growth and cultivation environment monitoring method |
| CN117807548B (en) * | 2024-02-29 | 2024-05-10 | 江苏新希望生态科技有限公司 | Bean sprout growth and cultivation environment monitoring method |
| CN118037716A (en) * | 2024-04-11 | 2024-05-14 | 贵州省园艺研究所(贵州省园艺工程技术研究中心) | Cowpea pest and disease damage identification method and system for field scene |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115456960A (en) | Citrus huanglongbing disease and pest monitoring and early warning system and method | |
| Ali et al. | Non-destructive techniques of detecting plant diseases: A review | |
| Zhao et al. | Detecting tomatoes in greenhouse scenes by combining AdaBoost classifier and colour analysis | |
| Lacar et al. | Use of hyperspectral imagery for mapping grape varieties in the Barossa Valley, South Australia | |
| Kiani et al. | Identification of plant disease infection using soft-computing: Application to modern botany | |
| US10729117B2 (en) | Pest monitoring method based on machine vision | |
| Lorente et al. | Early decay detection in citrus fruit using laser-light backscattering imaging | |
| CN107576618A (en) | Pyricularia Oryzae detection method and system based on depth convolutional neural networks | |
| CN107346434A (en) | A kind of plant pest detection method based on multiple features and SVMs | |
| Narvekar et al. | Grape leaf diseases detection & analysis using SGDM matrix method | |
| CN103065149A (en) | Netted melon fruit phenotype extraction and quantization method | |
| Ji et al. | In-field automatic detection of maize tassels using computer vision | |
| KR102297913B1 (en) | Plant growth monitoring system using hyperspectral reflected light and fluorescence scattering, and method thereof | |
| Lin et al. | A review on computer vision technologies applied in greenhouse plant stress detection | |
| CN114359539B (en) | Intelligent identification method for high-spectrum image of parasite in sashimi | |
| CN103049739A (en) | Tree detection method for use in intelligent monitoring of power transmission line | |
| Lang et al. | Detection of chlorophyll content in maize canopy from UAV imagery | |
| CN105067532A (en) | Method for identifying early-stage disease spots of sclerotinia sclerotiorum and botrytis of rape | |
| CN114199880A (en) | Citrus disease and insect pest real-time detection method based on edge calculation | |
| CN112106747A (en) | Intelligent agricultural insect pest remote automatic monitoring system | |
| CN113657294A (en) | Crop disease and insect pest detection method and system based on computer vision | |
| Yang et al. | Cotton hail disaster classification based on drone multispectral images at the flowering and boll stage | |
| CN115601670A (en) | Pine wilt disease monitoring method based on artificial intelligence and high-resolution remote sensing image | |
| Li et al. | Classification of pine wilt disease at different infection stages by diagnostic hyperspectral bands | |
| CN102184389B (en) | Automatic annual ring identification method used for tree species with high chromatic aberration of heartwood and sapwood |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |