WO2018120634A1 - Method and apparatus for identifying disease and insect damage - Google Patents

Method and apparatus for identifying disease and insect damage Download PDF

Info

Publication number
WO2018120634A1
WO2018120634A1 PCT/CN2017/086428 CN2017086428W WO2018120634A1 WO 2018120634 A1 WO2018120634 A1 WO 2018120634A1 CN 2017086428 W CN2017086428 W CN 2017086428W WO 2018120634 A1 WO2018120634 A1 WO 2018120634A1
Authority
WO
WIPO (PCT)
Prior art keywords
insect
pest
image data
disease
rgb image
Prior art date
Application number
PCT/CN2017/086428
Other languages
French (fr)
Chinese (zh)
Inventor
王刚
Original Assignee
深圳前海弘稼科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳前海弘稼科技有限公司 filed Critical 深圳前海弘稼科技有限公司
Publication of WO2018120634A1 publication Critical patent/WO2018120634A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features

Definitions

  • the invention relates to the field of agricultural technology, in particular to a pest and disease identification method, and to a pest and disease identification device.
  • the present invention aims to solve at least one of the technical problems existing in the prior art or related art.
  • an object of the present invention is to provide a method for identifying pests and diseases.
  • Another object of the present invention is to provide a pest identification device.
  • the present invention provides a method for identifying pests and diseases, comprising: collecting pest and disease images, converting pest and disease images into RGB image data; using unsupervised clustering algorithm to cluster and analyze RGB image data to obtain image classification of pests and diseases; The classified RGB image data is used for worm eye statistics to determine the number of insect spots; whether the number of insect spots is greater than the preset number of insect spots; when the judgment result is yes, an alarm prompt is issued.
  • the method for identifying pests and diseases of the present invention by collecting images of pests and diseases, preferably, by The camera takes a photo of the sticky board and parses the photo into pixel data in RGB format (RGB format is a method of encoding colors, collectively referred to as "color space” or “gamut”), using an unsupervised clustering algorithm. Clustering analysis of these pixel data (Clustering analysis is a typical application of unsupervised learning, and is a common method in exploratory data mining. Simply put, similar things are divided. Go to a group), get the classification of pests and diseases, and perform the eye-eye statistics on the classified data to confirm how many insect spots are on this image.
  • RGB format is a method of encoding colors, collectively referred to as "color space” or "gamut”
  • the clustering analysis is performed on the RGB image data by using an unsupervised clustering algorithm to obtain the step of classifying the pest image, specifically comprising: performing unsupervised clustering learning on the RGB image data, and establishing a similarity model; According to the similarity model, the RGB image data is classified and aggregated.
  • the unsupervised clustering algorithm is trained and learned, thereby establishing an optimal similarity model, and the RGB image data is aggregated and classified according to the similarity model to obtain a band.
  • the step of performing worm count on the classified RGB image data to determine the number of insect spots comprises: counting the number of worm eyes; calculating a ratio of the number of worm eyes to a preset threshold, according to The ratio determines the number of insects; when the ratio is an integer value, the integer value is used as the number of insects; when the ratio is a decimal, the decimal is rounded, and the rounded integer value is used as the number of insects.
  • the RGB image data of the classified worm is subjected to worm statistics, and preferably, a set of RGB image data with a worm eye is counted, and based on the number of the worm eyes, the number of the worm eyes and the preset threshold are calculated.
  • the ratio thereby determining the number of insect points based on the ratio.
  • the ratio is In the case of an integer, the integer value is used as the number of insects.
  • the ratio is a decimal, the decimal is rounded up and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
  • the preset threshold is 10.
  • the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point.
  • the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
  • the unsupervised clustering algorithm is K-means clustering.
  • K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.).
  • unsupervised clustering algorithm for cluster analysis other unsupervised clustering algorithms other than K-means clustering can also be selected.
  • the pest image classification includes: a worm eye data group and a non-worm eye data group.
  • the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
  • the invention also provides a pest and disease identification device, comprising: an image acquisition and processing unit for collecting pest and disease images, converting pest and disease images into RGB image data; and a pest and disease recognition unit for performing RGB image data using an unsupervised clustering algorithm.
  • Cluster analysis to obtain image classification of pests and diseases; statistical unit, used to perform worm statistics on the classified RGB image data to determine the number of insect points; the judgment unit is used to determine whether the number of insect spots is greater than the number of preset pests; It is used to issue an alarm when the judgment result is yes.
  • the camera takes a photo of the sticky board and parses the photo into pixel data in RGB format (RGB format is a method of encoding colors, collectively referred to as "color space” or “gamut”), using an unsupervised clustering algorithm.
  • RGB format is a method of encoding colors, collectively referred to as "color space” or "gamut”
  • Clustering analysis is a typical application of unsupervised learning, and is a common method in exploratory data mining. Simply put, similar things are divided. Go to a group), get the classification of pests and diseases, and perform the eye-eye statistics on the classified data to confirm how many insect spots are on this image. When the number of insect spots exceeds the preset value, it indicates that there are more insect spots and serious pests and diseases.
  • the pest identification unit performs cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a step of classifying the pest image, specifically comprising: a modeling unit for unsupervising the RGB image data. Clustering learning, establishing a similarity model; a classification unit for classifying RGB image data according to the similarity model.
  • the unsupervised clustering algorithm is trained and learned, thereby establishing an optimal similarity model, and the RGB image data is aggregated and classified according to the similarity model to obtain a band.
  • the step of performing statistics on the RGB image data of the classified statistic unit, and determining the number of insect points specifically comprising: a counting unit for counting the number of insect eyes; and a calculating unit, configured to: Calculate the ratio of the number of insect eyes to the preset threshold, and determine the number of insect points according to the ratio; the calculation unit is specifically used to use the integer value as the number of insect points when the ratio is an integer value; the calculation unit is specifically used when the ratio is In the case of a decimal, the decimal is rounded up and the rounded integer value is used as the number of insects.
  • the RGB image data of the classified worm is subjected to worm statistics, and preferably, a set of RGB image data with a worm eye is counted, and based on the number of the worm eyes, the number of the worm eyes and the preset threshold are calculated.
  • the ratio thereby determining the number of insect points based on the ratio.
  • the ratio is an integer
  • the integer value is used as the number of insects.
  • the ratio is a decimal
  • the decimal is rounded up. The rounded value is used as the number of insect points. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
  • the preset threshold is 10.
  • the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point.
  • the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
  • the unsupervised clustering algorithm is K-means clustering.
  • K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.).
  • unsupervised clustering algorithm for cluster analysis other unsupervised clustering algorithms other than K-means clustering can also be selected.
  • the pest image classification includes: a worm eye data group and a non-worm eye data group.
  • the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
  • FIG. 1 is a flow chart showing a method for identifying pests and diseases according to an embodiment of the present invention
  • FIG. 2 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention.
  • FIG. 3 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention.
  • FIG. 4 is a schematic block diagram of a pest and disease recognition apparatus according to an embodiment of the present invention.
  • Figure 5 is a schematic block diagram of a pest and disease recognition apparatus according to still another embodiment of the present invention.
  • Fig. 6 shows a schematic block diagram of a pest identification device according to still another embodiment of the present invention.
  • FIG. 1 is a flow chart showing a method for identifying pests and diseases according to an embodiment of the present invention. Among them, the method for identifying pests and diseases includes:
  • Step 102 collecting a pest and disease image, and converting the pest image into RGB image data
  • Step 104 Perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification
  • Step 106 Perform worm statistics on the classified RGB image data to determine the number of insect spots
  • Step 108 Determine whether the number of insect points is greater than a preset number of insect points
  • step 110 when the judgment result is yes, an alarm prompt is issued.
  • the camera photographs the sticky insect board and parses the photo into an RGB format (the RGB format is a method of encoding the color, collectively referred to as “color space” or “ Pixel data of the gamut”, using unsupervised clustering algorithm to cluster these pixel data (clustering analysis is a typical application of unsupervised learning, and also in exploratory data mining
  • clustering analysis is a typical application of unsupervised learning, and also in exploratory data mining
  • a common method which is simply to group similar things into groups, to get the classification of pests and diseases, and to perform statistics on the classified data, it is possible to confirm how many insect spots are on this image. When the number exceeds the preset value, it indicates that there are more insects, serious pests and diseases, and an alarm is issued.
  • effective identification of agricultural pests and diseases and timely reminding relevant personnel to carry out diseases Pest control avoiding problems such as agricultural production reduction due to pests and diseases, declining quality of agricultural products, and economic losses.
  • the method for identifying pests and diseases includes:
  • Step 202 collecting a pest and disease image, and converting the pest image into RGB image data
  • the clustering analysis of RGB image data is performed by using an unsupervised clustering algorithm to obtain the steps of classifying pest and disease images, including:
  • Step 204 Perform unsupervised clustering learning on the RGB image data to establish a similarity model.
  • Step 206 Perform aggregation classification on the RGB image data according to the similarity model.
  • Step 208 Perform worm statistics on the classified RGB image data to determine the number of insect spots
  • Step 210 Determine whether the number of insect points is greater than a preset number of insect points
  • Step 212 When the judgment result is yes, an alarm prompt is issued.
  • the unsupervised clustering algorithm is trained and learned to establish an optimal similarity model, and the RGB image data is classified and classified according to the similarity model to obtain a band.
  • FIG. 3 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention.
  • the method for identifying pests and diseases includes:
  • Step 302 collecting a pest and disease image, and converting the pest image into RGB image data
  • the clustering analysis of RGB image data is performed by using an unsupervised clustering algorithm to obtain the steps of classifying pest and disease images, including:
  • Step 304 Perform unsupervised clustering learning on the RGB image data to establish a similarity model
  • Step 306 Perform aggregation classification on the RGB image data according to the similarity model.
  • the step of performing worm count statistics on the classified RGB image data to determine the number of insect spots includes:
  • Step 308 counting the number of insect eyes
  • Step 310 Calculate a ratio of the number of insect eyes to a preset threshold, and determine the number of insect points according to the ratio;
  • Step 312 when the ratio is an integer value, the integer value is used as the number of insect points; when the ratio is a decimal, the decimal carry is rounded, and the rounded integer value is used as the number of insect points;
  • Step 314 determining whether the number of insect points is greater than a preset number of insect points
  • step 316 when the judgment result is yes, an alarm prompt is issued.
  • performing worm statistics on the classified RGB image data preferably, counting a set of RGB image data with a wormhole, and calculating the number of wormholes and a preset threshold based on the number of wormholes obtained.
  • the ratio thereby determining the number of insect points based on the ratio.
  • the ratio is an integer
  • the integer value is used as the number of insects.
  • the ratio is a decimal
  • the decimal is rounded up, and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
  • the preset threshold is 10.
  • the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point.
  • the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
  • the unsupervised clustering algorithm is K-means clustering.
  • K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.).
  • unsupervised clustering algorithm for cluster analysis other unsupervised clustering algorithms other than K-means clustering can also be selected.
  • the pest image classification includes: a worm eye data set and a non-worm eye data set.
  • the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
  • the pest identification device 400 includes:
  • the image acquisition and processing unit 402 is configured to collect a pest image and convert the pest image into RGB image data.
  • the pest and disease identification unit 404 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
  • the statistic unit 406 is configured to perform worm statistics on the classified RGB image data to determine the number of insect spots
  • the determining unit 408 is configured to determine whether the number of insects is greater than a preset number of insects
  • the reminding unit 410 is configured to issue an alarm prompt when the determination result is yes.
  • the photo of the sticky board is photographed by the camera, and the photo is parsed into an RGB format (the RGB format is a method of encoding the color, collectively referred to as "color space” or " Pixel data of the gamut", using unsupervised clustering algorithm to cluster these pixel data (clustering analysis is a typical application of unsupervised learning, and also in exploratory data mining A common method, which is simply to group similar things into groups, to get the classification of pests and diseases, and to perform statistics on the classified data, it is possible to confirm how many insect spots are on this image. When the number exceeds the preset value, it indicates that there are more insects, serious pests and diseases, and an alarm is issued.
  • agricultural pests and diseases are effectively identified, and relevant personnel are promptly reminded to carry out pest control, and problems such as agricultural production reduction, agricultural product quality degradation, and economic loss caused by pests and diseases are avoided.
  • the pest identification device 500 includes:
  • the image acquisition and processing unit 502 is configured to collect a pest image and convert the pest image into RGB image data.
  • the pest and disease identification unit 504 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
  • the statistic unit 506 is configured to perform worm statistics on the classified RGB image data to determine the number of insect spots;
  • the determining unit 508 is configured to determine whether the number of insects is greater than a preset number of insects
  • the reminding unit 510 is configured to issue an alarm prompt when the determination result is yes;
  • the pest identification unit 504 specifically includes:
  • the modeling unit 5042 is configured to perform unsupervised clustering learning on the RGB image data to establish a similarity model
  • the classification unit 5044 is configured to perform aggregation and classification on the RGB image data according to the similarity model.
  • the unsupervised clustering algorithm is trained and learned to establish an optimal similarity model, and the RGB image data is classified and classified according to the similarity model to obtain a band.
  • the pest identification device 600 includes:
  • An image acquisition and processing unit 602 is configured to collect a pest image and convert the pest image into RGB image data
  • the pest and disease identification unit 604 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
  • the statistic unit 606 is configured to perform worm statistics on the classified RGB image data to determine the number of insect points
  • the determining unit 608 is configured to determine whether the number of insect points is greater than a preset number of insect points
  • the reminding unit 610 is configured to issue an alarm prompt when the determination result is yes;
  • the pest and disease identification unit 604 specifically includes:
  • the modeling unit 6042 is configured to perform unsupervised clustering learning on the RGB image data to establish a similarity model
  • the classification unit 6044 is configured to perform aggregation and classification on the RGB image data according to the similarity model
  • the statistics unit 606 specifically includes:
  • a counting unit 6062 configured to count the number of insect eyes
  • a calculating unit 6064 configured to calculate a ratio of the number of insect eyes to a preset threshold, and determine a number of insect points according to the ratio
  • a calculating unit 6062 specifically, when the ratio is an integer value, The integer value is used as the number of bugs; the calculating unit 6064 is specifically used to round the decimals when the ratio is a decimal, and use the integer value as the number of insects. Measured value.
  • performing worm statistics on the classified RGB image data preferably, counting a set of RGB image data with a wormhole, and calculating the number of wormholes and a preset threshold based on the number of wormholes obtained.
  • the ratio thereby determining the number of insect points based on the ratio.
  • the ratio is an integer
  • the integer value is used as the number of insects.
  • the ratio is a decimal
  • the decimal is rounded up, and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
  • the preset threshold is 10.
  • the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point.
  • the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
  • the unsupervised clustering algorithm is K-means clustering.
  • K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.).
  • unsupervised clustering algorithm for cluster analysis other unsupervised clustering algorithms other than K-means clustering can also be selected.
  • the pest image classification includes: a worm eye data set and a non-worm eye data set.
  • the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
  • a photo of the vegetable is collected by the camera, and the photo is parsed into RGB data, and the RGB data is unsupervised clustered (K-means), and the RGB pixel data is aggregated into a group according to the similarity.
  • K-means unsupervised clustered
  • the captured photos are converted into a picture as RGB values.
  • unsupervised clustering learning is performed and a similarity model is established.
  • the RGB values are generated into two categories, one is the RGB data of the insect eye, and the other is It is non-worm eye RGB data, in which the RGB data category of the insect eye is 0, and the RGB data of the non-worm eye is 1; finally, the data of the insect eye is counted, specifically, the number of the statistical category is 0, and the measured 10 pixel is a pest point. It is calculated that there are 4 insect spots on this map.
  • the description of the terms “one embodiment”, “some embodiments”, “specific embodiments” and the like means that the specific features, structures, materials, or characteristics described in connection with the embodiments or examples are included in the present invention. At least one embodiment or example.
  • the schematic representation of the above terms does not necessarily refer to the same embodiment or example.
  • the particular features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

A method and apparatus for identifying a disease and an insect damage. The method for identifying a disease and an insect damage comprises: collecting a disease and insect damage image, and converting the disease and insect damage image into RGB image data (102); using an unsupervised clustering algorithm to perform clustering analysis on the RGB image data to obtain the classification of the disease and insect damage image (104); performing insect hole statistics on the classified RGB image data to determine the number of insect points (106); determining whether the number of insect points is greater than the pre-set number of insect points (108); and when a determination result is yes, sending out an alarm prompt (110). By means of the method, agricultural diseases and insect damages are effectively identified, and relevant persons are alerted in time to prevent and treat diseases and insect damages, thereby preventing the problems of a reduction in agricultural production, a reduction in agricultural product quality, economic losses, etc. caused by the diseases and insect damages.

Description

病虫害识别方法及装置Disease and pest identification method and device
本申请要求于2016年12月30日提交中国专利局、申请号为201611259656.0发明名称为“病虫害识别方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。The present application claims priority to Chinese Patent Application No. Serial No. No. No. No. No. No. No. No. No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No
技术领域Technical field
本发明涉及农业技术领域，具体而言，涉及一种病虫害识别方法，还涉及一种病虫害识别装置。The invention relates to the field of agricultural technology, in particular to a pest and disease identification method, and to a pest and disease identification device.
背景技术Background technique
在农业的可持续发展中，病虫害的识别与防治起到了非常重要的作用。农业病虫害对农业生产安全、人们身体健康及环境安全有着直接的影响，不仅能够导致农业减产，农产品质量下降，更会造成农民收入。然而，相关农业云技术中没有农业病虫害图像识别方案。In the sustainable development of agriculture, the identification and prevention of pests and diseases plays a very important role. Agricultural pests and diseases have a direct impact on agricultural production safety, people's health and environmental safety, which can not only lead to agricultural production reduction, but also the quality of agricultural products, and will also cause farmers' income. However, there is no agricultural pest and disease image recognition scheme in the relevant agricultural cloud technology.
因此，如何提供一种有效的病虫害识别方法，成为目前亟待解决的技术问题。Therefore, how to provide an effective pest and disease identification method has become a technical problem that needs to be solved urgently.
发明内容Summary of the invention
本发明旨在至少解决现有技术或相关技术中存在的技术问题之一。The present invention aims to solve at least one of the technical problems existing in the prior art or related art.
为此，本发明的一个目的在于提出了一种病虫害识别方法。To this end, an object of the present invention is to provide a method for identifying pests and diseases.
本发明的另一个目的在于提出了一种病虫害识别装置。Another object of the present invention is to provide a pest identification device.
有鉴于此，本发明提出了一种病虫害识别方法，包括：采集病虫害图像，将病虫害图像转换成RGB图像数据；使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；对分类后的RGB图像数据进行虫眼统计，确定虫点数量；判断虫点数量是否大于预设虫点数量；当判断结果为是时，发出报警提示。In view of this, the present invention provides a method for identifying pests and diseases, comprising: collecting pest and disease images, converting pest and disease images into RGB image data; using unsupervised clustering algorithm to cluster and analyze RGB image data to obtain image classification of pests and diseases; The classified RGB image data is used for worm eye statistics to determine the number of insect spots; whether the number of insect spots is greater than the preset number of insect spots; when the judgment result is yes, an alarm prompt is issued.
根据本发明的病虫害识别方法，通过采集病虫害图像，优选的，通过 摄像头拍摄粘虫板照片，并将照片解析成RGB格式(RGB格式为一种对颜色进行编码的方法，统称为“颜色空间”或“色域”)的像素数据，使用无监督聚类算法对这些像素数据进行聚类分析(聚类(Clustering)分析是无监督式机器学习(unsupervised learning)的一个典型应用，也是探索性数据挖掘中的一种常用方法，简单地说就是把相似的东西分到一组)，得到病虫害的分类，对分类后的数据进行虫眼统计，就能够确认这张图像上有多少虫点，当虫点的数量超过预设数值时，说明虫点较多，病虫害严重，发出报警提示。通过本发明的技术方案，有效识别农业病虫害并及时的提醒相关人员进行病虫害防治，避免由于病虫害导致的农业减产、农产品质量下降、经济损失等问题。According to the method for identifying pests and diseases of the present invention, by collecting images of pests and diseases, preferably, by The camera takes a photo of the sticky board and parses the photo into pixel data in RGB format (RGB format is a method of encoding colors, collectively referred to as "color space" or "gamut"), using an unsupervised clustering algorithm. Clustering analysis of these pixel data (Clustering analysis is a typical application of unsupervised learning, and is a common method in exploratory data mining. Simply put, similar things are divided. Go to a group), get the classification of pests and diseases, and perform the eye-eye statistics on the classified data to confirm how many insect spots are on this image. When the number of insect spots exceeds the preset value, it indicates that there are more insect spots and serious pests and diseases. , issued an alarm prompt. Through the technical scheme of the invention, the agricultural pests and diseases are effectively identified, and relevant personnel are promptly reminded to carry out pest control, and problems such as agricultural production reduction, quality deterioration of agricultural products and economic loss caused by pests and diseases are avoided.
另外，根据本发明上述的病虫害识别方法，还可以具有如下附加的技术特征：Further, according to the above-described pest identification method of the present invention, it is also possible to have the following additional technical features:
在上述技术方案中，优选地，使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：对RGB图像数据进行无监督聚类学习，建立相似度模型；根据相似度模型，对RGB图像数据进行聚合分类。In the above technical solution, preferably, the clustering analysis is performed on the RGB image data by using an unsupervised clustering algorithm to obtain the step of classifying the pest image, specifically comprising: performing unsupervised clustering learning on the RGB image data, and establishing a similarity model; According to the similarity model, the RGB image data is classified and aggregated.
在该技术方案中，将病虫害图像解析为RGB图像数据后，通过对无监督聚类算法进行训练学习，从而建立最佳相似度模型，根据相似度模型，对RGB图像数据进行聚合分类，得到带虫眼的数据组以及不带虫眼的数据组，并在分类的基础上通过对带虫眼数据的统计分析，准确确认虫点个数，从而有效识别病虫害图像。In the technical solution, after the pest and disease image is parsed into RGB image data, the unsupervised clustering algorithm is trained and learned, thereby establishing an optimal similarity model, and the RGB image data is aggregated and classified according to the similarity model to obtain a band. The data set of the insect eye and the data set without the insect eye, and the statistical analysis of the data of the insect eye on the basis of the classification, accurately confirm the number of insect spots, thereby effectively identifying the pest and disease image.
在上述任一技术方案中，优选地，对分类后的RGB图像数据的进行虫眼统计，确定虫点数量的步骤，具体包括：统计虫眼的数量；计算虫眼的数量与预设阈值的比值，根据比值确定虫点数量；当比值为整数值时，将整数值作为虫点数量值；当比值为小数时，将小数进位取整，将取整后的整数值作为虫点数量值。In any one of the above technical solutions, preferably, the step of performing worm count on the classified RGB image data to determine the number of insect spots comprises: counting the number of worm eyes; calculating a ratio of the number of worm eyes to a preset threshold, according to The ratio determines the number of insects; when the ratio is an integer value, the integer value is used as the number of insects; when the ratio is a decimal, the decimal is rounded, and the rounded integer value is used as the number of insects.
在该技术方案中，对分类后的RGB图像数据的进行虫眼统计，优选的，对带有虫眼的一组RGB图像数据进行统计，在得到虫眼数量的基础上，计算虫眼数量与预设阈值的比值，从而根据比值确定虫点的数量。当比值为 整数时，将该整数值作为虫点数，当比值为小数时，将小数进位取整，将取整后的数值作为虫点数。举例来说，统计虫眼点的个数为35，预设阈值为10，那么可以判定这张图像上有4个虫点。In the technical solution, the RGB image data of the classified worm is subjected to worm statistics, and preferably, a set of RGB image data with a worm eye is counted, and based on the number of the worm eyes, the number of the worm eyes and the preset threshold are calculated. The ratio, thereby determining the number of insect points based on the ratio. When the ratio is In the case of an integer, the integer value is used as the number of insects. When the ratio is a decimal, the decimal is rounded up and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
在上述任一技术方案中，优选地，预设阈值为10。In any of the above technical solutions, preferably, the preset threshold is 10.
在该技术方案中，预设阈值为10，说明10个RGB虫眼数据构成一个虫点。本领域技术人员应该理解，预设阈值为10但不限于此，由于不同作物的种植收到多种因素的影响，如地域、季节、土壤条件、气候等，因而产生的病虫害也会大有不同，从而经过实测统计后的预设阈值也会相应变化。In this technical solution, the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point. Those skilled in the art should understand that the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
在上述任一技术方案中，优选地，无监督聚类算法为K均值聚类。In any of the above technical solutions, preferably, the unsupervised clustering algorithm is K-means clustering.
在该技术方案中，K均值聚类(K-means clustering)是最典型的聚类算法(当然，除此之外，还有很多诸如属于划分法K-MEDOIDS算法、CLARANS算法；属于层次法的BIRCH算法、CURE算法、CHAMELEON算法等；基于密度的方法：DBSCAN算法、OPTICS算法、DENCLUE算法等；基于网格的方法：STING算法、CLIQUE算法、WAVE-CLUSTER算法；基于模型的方法等)。本领域技术人员应该理解，在使用无监督聚类算法进行聚类分析时，也可以选择K均值聚类以外的其它无监督聚类算法。In this technical solution, K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.). Those skilled in the art should understand that when using the unsupervised clustering algorithm for cluster analysis, other unsupervised clustering algorithms other than K-means clustering can also be selected.
在上述任一技术方案中，优选地，病虫害图像分类包括：虫眼数据组和非虫眼数据组。In any of the above technical solutions, preferably, the pest image classification includes: a worm eye data group and a non-worm eye data group.
在该技术方案中，通过无监督聚类算法对病虫害图像进行聚类分析后，得到虫眼数据组和非虫眼数据组，从而只需对虫眼数据组进行统计，以确认虫点的个数，实现病虫害的有效识别。In the technical solution, after the cluster analysis of the pest and disease image by the unsupervised clustering algorithm, the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
本发明还提出一种病虫害识别装置，包括：图像采集与处理单元，用于采集病虫害图像，将病虫害图像转换成RGB图像数据；病虫害识别单元，用于使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；统计单元，用于对分类后的RGB图像数据进行虫眼统计，确定虫点数量；判断单元，用于判断虫点数量是否大于预设虫点数量；提醒单元，用于当判断结果为是时，发出报警提示。The invention also provides a pest and disease identification device, comprising: an image acquisition and processing unit for collecting pest and disease images, converting pest and disease images into RGB image data; and a pest and disease recognition unit for performing RGB image data using an unsupervised clustering algorithm. Cluster analysis, to obtain image classification of pests and diseases; statistical unit, used to perform worm statistics on the classified RGB image data to determine the number of insect points; the judgment unit is used to determine whether the number of insect spots is greater than the number of preset pests; It is used to issue an alarm when the judgment result is yes.
根据本发明的病虫害识别装置，通过采集病虫害图像，优选的，通过 摄像头拍摄粘虫板照片，并将照片解析成RGB格式(RGB格式为一种对颜色进行编码的方法，统称为“颜色空间”或“色域”)的像素数据，使用无监督聚类算法对这些像素数据进行聚类分析(聚类(Clustering)分析是无监督式机器学习(unsupervised learning)的一个典型应用，也是探索性数据挖掘中的一种常用方法，简单地说就是把相似的东西分到一组)，得到病虫害的分类，对分类后的数据进行虫眼统计，就能够确认这张图像上有多少虫点，当虫点的数量超过预设数值时，说明虫点较多，病虫害严重，发出报警提示。通过本发明的技术方案，有效识别农业病虫害并及时的提醒相关人员进行病虫害防治，避免由于病虫害导致的农业减产、农产品质量下降、经济损失等问题。According to the pest identification device of the present invention, by collecting images of pests and diseases, preferably, The camera takes a photo of the sticky board and parses the photo into pixel data in RGB format (RGB format is a method of encoding colors, collectively referred to as "color space" or "gamut"), using an unsupervised clustering algorithm. Clustering analysis of these pixel data (Clustering analysis is a typical application of unsupervised learning, and is a common method in exploratory data mining. Simply put, similar things are divided. Go to a group), get the classification of pests and diseases, and perform the eye-eye statistics on the classified data to confirm how many insect spots are on this image. When the number of insect spots exceeds the preset value, it indicates that there are more insect spots and serious pests and diseases. , issued an alarm prompt. Through the technical scheme of the invention, the agricultural pests and diseases are effectively identified, and relevant personnel are promptly reminded to carry out pest control, and problems such as agricultural production reduction, quality deterioration of agricultural products and economic loss caused by pests and diseases are avoided.
在上述技术方案中，优选地，病虫害识别单元使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：建模单元，用于对RGB图像数据进行无监督聚类学习，建立相似度模型；分类单元，用于根据相似度模型，对RGB图像数据进行聚合分类。In the above technical solution, preferably, the pest identification unit performs cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a step of classifying the pest image, specifically comprising: a modeling unit for unsupervising the RGB image data. Clustering learning, establishing a similarity model; a classification unit for classifying RGB image data according to the similarity model.
在该技术方案中，将病虫害图像解析为RGB图像数据后，通过对无监督聚类算法进行训练学习，从而建立最佳相似度模型，根据相似度模型，对RGB图像数据进行聚合分类，得到带虫眼的数据组以及不带虫眼的数据组，并在分类的基础上通过对带虫眼数据的统计分析，准确确认虫点个数，从而有效识别病虫害图像。In the technical solution, after the pest and disease image is parsed into RGB image data, the unsupervised clustering algorithm is trained and learned, thereby establishing an optimal similarity model, and the RGB image data is aggregated and classified according to the similarity model to obtain a band. The data set of the insect eye and the data set without the insect eye, and the statistical analysis of the data of the insect eye on the basis of the classification, accurately confirm the number of insect spots, thereby effectively identifying the pest and disease image.
在上述任一技术方案中，优选地，统计单元对分类后的RGB图像数据的进行虫眼统计，确定虫点数量的步骤，具体包括：计数单元，用于统计虫眼的数量；计算单元，用于计算虫眼的数量与预设阈值的比值，根据比值确定虫点数量；计算单元，具体用于当比值为整数值时，将整数值作为虫点数量值；计算单元，具体还用于当比值为小数时，将小数进位取整，将取整后的整数值作为虫点数量值。In any one of the above technical solutions, preferably, the step of performing statistics on the RGB image data of the classified statistic unit, and determining the number of insect points, specifically comprising: a counting unit for counting the number of insect eyes; and a calculating unit, configured to: Calculate the ratio of the number of insect eyes to the preset threshold, and determine the number of insect points according to the ratio; the calculation unit is specifically used to use the integer value as the number of insect points when the ratio is an integer value; the calculation unit is specifically used when the ratio is In the case of a decimal, the decimal is rounded up and the rounded integer value is used as the number of insects.
在该技术方案中，对分类后的RGB图像数据的进行虫眼统计，优选的，对带有虫眼的一组RGB图像数据进行统计，在得到虫眼数量的基础上，计算虫眼数量与预设阈值的比值，从而根据比值确定虫点的数量。当比值为整数时，将该整数值作为虫点数，当比值为小数时，将小数进位取整，将 取整后的数值作为虫点数。举例来说，统计虫眼点的个数为35，预设阈值为10，那么可以判定这张图像上有4个虫点。In the technical solution, the RGB image data of the classified worm is subjected to worm statistics, and preferably, a set of RGB image data with a worm eye is counted, and based on the number of the worm eyes, the number of the worm eyes and the preset threshold are calculated. The ratio, thereby determining the number of insect points based on the ratio. When the ratio is an integer, the integer value is used as the number of insects. When the ratio is a decimal, the decimal is rounded up. The rounded value is used as the number of insect points. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
在上述任一技术方案中，优选地，预设阈值为10。In any of the above technical solutions, preferably, the preset threshold is 10.
在该技术方案中，预设阈值为10，说明10个RGB虫眼数据构成一个虫点。本领域技术人员应该理解，预设阈值为10但不限于此，由于不同作物的种植收到多种因素的影响，如地域、季节、土壤条件、气候等，因而产生的病虫害也会大有不同，从而经过实测统计后的预设阈值也会相应变化。In this technical solution, the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point. Those skilled in the art should understand that the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
在上述任一技术方案中，优选地，无监督聚类算法为K均值聚类。In any of the above technical solutions, preferably, the unsupervised clustering algorithm is K-means clustering.
在该技术方案中，K均值聚类(K-means clustering)是最典型的聚类算法(当然，除此之外，还有很多诸如属于划分法K-MEDOIDS算法、CLARANS算法；属于层次法的BIRCH算法、CURE算法、CHAMELEON算法等；基于密度的方法：DBSCAN算法、OPTICS算法、DENCLUE算法等；基于网格的方法：STING算法、CLIQUE算法、WAVE-CLUSTER算法；基于模型的方法等)。本领域技术人员应该理解，在使用无监督聚类算法进行聚类分析时，也可以选择K均值聚类以外的其它无监督聚类算法。In this technical solution, K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.). Those skilled in the art should understand that when using the unsupervised clustering algorithm for cluster analysis, other unsupervised clustering algorithms other than K-means clustering can also be selected.
在上述任一技术方案中，优选地，病虫害图像分类包括：虫眼数据组和非虫眼数据组。In any of the above technical solutions, preferably, the pest image classification includes: a worm eye data group and a non-worm eye data group.
在该技术方案中，通过无监督聚类算法对病虫害图像进行聚类分析后，得到虫眼数据组和非虫眼数据组，从而只需对虫眼数据组进行统计，以确认虫点的个数，实现病虫害的有效识别。In the technical solution, after the cluster analysis of the pest and disease image by the unsupervised clustering algorithm, the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
本发明的附加方面和优点将在下面的描述部分中变得明显，或通过本发明的实践了解到。Additional aspects and advantages of the invention will be apparent from the description of the invention.
附图说明DRAWINGS
本发明的上述和/或附加的方面和优点从结合下面附图对实施例的描述中将变得明显和容易理解，其中：The above and/or additional aspects and advantages of the present invention will become apparent and readily understood from
图1示出了根据本发明一实施例的病虫害识别方法的流程示意图；1 is a flow chart showing a method for identifying pests and diseases according to an embodiment of the present invention;
图2示出了根据本发明再一实施例的病虫害识别方法的流程示意图； 2 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention;
图3示出了根据本发明又一实施例的病虫害识别方法的流程示意图；3 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention;
图4示出了根据本发明一实施例的病虫害识别装置的示意框图；4 is a schematic block diagram of a pest and disease recognition apparatus according to an embodiment of the present invention;
图5示出了根据本发明再一实施例的病虫害识别装置的示意框图；Figure 5 is a schematic block diagram of a pest and disease recognition apparatus according to still another embodiment of the present invention;
图6示出了根据本发明又一实施例的病虫害识别装置的示意框图。Fig. 6 shows a schematic block diagram of a pest identification device according to still another embodiment of the present invention.
具体实施方式detailed description
为了能够更清楚地理解本发明的上述目的、特征和优点，下面结合附图和具体实施方式对本发明进行进一步的详细描述。需要说明的是，在不冲突的情况下，本申请的实施例及实施例中的特征可以相互组合。The present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that the embodiments in the present application and the features in the embodiments may be combined with each other without conflict.
在下面的描述中阐述了很多具体细节以便于充分理解本发明，但是，本发明还可以采用其他不同于在此描述的其他方式来实施，因此，本发明的保护范围并不受下面公开的具体实施例的限制。In the following description, numerous specific details are set forth in order to provide a full understanding of the invention, but the invention may be practiced otherwise than as described herein. Limitations of the embodiments.
如图1所示，根据本发明的一个实施例的病虫害识别方法的流程示意图。其中，该病虫害识别方法包括：FIG. 1 is a flow chart showing a method for identifying pests and diseases according to an embodiment of the present invention. Among them, the method for identifying pests and diseases includes:
步骤102，采集病虫害图像，将病虫害图像转换成RGB图像数据； Step 102, collecting a pest and disease image, and converting the pest image into RGB image data;
步骤104，使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；Step 104: Perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
步骤106，对分类后的RGB图像数据进行虫眼统计，确定虫点数量；Step 106: Perform worm statistics on the classified RGB image data to determine the number of insect spots;
步骤108，判断虫点数量是否大于预设虫点数量；Step 108: Determine whether the number of insect points is greater than a preset number of insect points;
步骤110，当判断结果为是时，发出报警提示。In step 110, when the judgment result is yes, an alarm prompt is issued.
在该技术方案中，通过采集病虫害图像，优选的，通过摄像头拍摄粘虫板照片，并将照片解析成RGB格式(RGB格式为一种对颜色进行编码的方法，统称为“颜色空间”或“色域”)的像素数据，使用无监督聚类算法对这些像素数据进行聚类分析(聚类(Clustering)分析是无监督式机器学习(unsupervised learning)的一个典型应用，也是探索性数据挖掘中的一种常用方法，简单地说就是把相似的东西分到一组)，得到病虫害的分类，对分类后的数据进行虫眼统计，就能够确认这张图像上有多少虫点，当虫点的数量超过预设数值时，说明虫点较多，病虫害严重，发出报警提示。通过本发明的实施例，有效识别农业病虫害并及时的提醒相关人员进行病 虫害防治，避免由于病虫害导致的农业减产、农产品质量下降、经济损失等问题。In the technical solution, by collecting the pest image, preferably, the camera photographs the sticky insect board and parses the photo into an RGB format (the RGB format is a method of encoding the color, collectively referred to as “color space” or “ Pixel data of the gamut", using unsupervised clustering algorithm to cluster these pixel data (clustering analysis is a typical application of unsupervised learning, and also in exploratory data mining A common method, which is simply to group similar things into groups, to get the classification of pests and diseases, and to perform statistics on the classified data, it is possible to confirm how many insect spots are on this image. When the number exceeds the preset value, it indicates that there are more insects, serious pests and diseases, and an alarm is issued. Through the embodiments of the present invention, effective identification of agricultural pests and diseases and timely reminding relevant personnel to carry out diseases Pest control, avoiding problems such as agricultural production reduction due to pests and diseases, declining quality of agricultural products, and economic losses.
如图2所示，根据本发明的再一个实施例的病虫害识别方法的流程示意图。其中，该病虫害识别方法包括：2 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention. Among them, the method for identifying pests and diseases includes:
步骤202，采集病虫害图像，将病虫害图像转换成RGB图像数据； Step 202, collecting a pest and disease image, and converting the pest image into RGB image data;
使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：The clustering analysis of RGB image data is performed by using an unsupervised clustering algorithm to obtain the steps of classifying pest and disease images, including:
步骤204，对RGB图像数据进行无监督聚类学习，建立相似度模型；Step 204: Perform unsupervised clustering learning on the RGB image data to establish a similarity model.
步骤206，根据相似度模型，对RGB图像数据进行聚合分类；Step 206: Perform aggregation classification on the RGB image data according to the similarity model.
步骤208，对分类后的RGB图像数据进行虫眼统计，确定虫点数量；Step 208: Perform worm statistics on the classified RGB image data to determine the number of insect spots;
步骤210，判断虫点数量是否大于预设虫点数量；Step 210: Determine whether the number of insect points is greater than a preset number of insect points;
步骤212，当判断结果为是时，发出报警提示。Step 212: When the judgment result is yes, an alarm prompt is issued.
在该实施例中，将病虫害图像解析为RGB图像数据后，通过对无监督聚类算法进行训练学习，从而建立最佳相似度模型，根据相似度模型，对RGB图像数据进行聚合分类，得到带虫眼的数据组以及不带虫眼的数据组，并在分类的基础上通过对带虫眼数据的统计分析，准确确认虫点个数，从而有效识别病虫害图像。In this embodiment, after the pest image is parsed into RGB image data, the unsupervised clustering algorithm is trained and learned to establish an optimal similarity model, and the RGB image data is classified and classified according to the similarity model to obtain a band. The data set of the insect eye and the data set without the insect eye, and the statistical analysis of the data of the insect eye on the basis of the classification, accurately confirm the number of insect spots, thereby effectively identifying the pest and disease image.
如图3所示，根据本发明的又一个实施例的病虫害识别方法的流程示意图。其中，该病虫害识别方法包括：FIG. 3 is a flow chart showing a method for identifying pests and diseases according to still another embodiment of the present invention. Among them, the method for identifying pests and diseases includes:
步骤302，采集病虫害图像，将病虫害图像转换成RGB图像数据； Step 302, collecting a pest and disease image, and converting the pest image into RGB image data;
使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：The clustering analysis of RGB image data is performed by using an unsupervised clustering algorithm to obtain the steps of classifying pest and disease images, including:
步骤304，对RGB图像数据进行无监督聚类学习，建立相似度模型；Step 304: Perform unsupervised clustering learning on the RGB image data to establish a similarity model;
步骤306，根据相似度模型，对RGB图像数据进行聚合分类；Step 306: Perform aggregation classification on the RGB image data according to the similarity model.
对分类后的RGB图像数据进行虫眼统计，确定虫点数量的步骤，具体包括：The step of performing worm count statistics on the classified RGB image data to determine the number of insect spots includes:
步骤308，统计虫眼的数量； Step 308, counting the number of insect eyes;
步骤310，计算虫眼的数量与预设阈值的比值，根据比值确定虫点数量； Step 310: Calculate a ratio of the number of insect eyes to a preset threshold, and determine the number of insect points according to the ratio;
步骤312，当比值为整数值时，将整数值作为虫点数量值；当比值为小数时，将小数进位取整，将取整后的整数值作为虫点数量值； Step 312, when the ratio is an integer value, the integer value is used as the number of insect points; when the ratio is a decimal, the decimal carry is rounded, and the rounded integer value is used as the number of insect points;
步骤314，判断虫点数量是否大于预设虫点数量； Step 314, determining whether the number of insect points is greater than a preset number of insect points;
步骤316，当判断结果为是时，发出报警提示。In step 316, when the judgment result is yes, an alarm prompt is issued.
在该实施例中，对分类后的RGB图像数据的进行虫眼统计，优选的，对带有虫眼的一组RGB图像数据进行统计，在得到虫眼数量的基础上，计算虫眼数量与预设阈值的比值，从而根据比值确定虫点的数量。当比值为整数时，将该整数值作为虫点数，当比值为小数时，将小数进位取整，将取整后的数值作为虫点数。举例来说，统计虫眼点的个数为35，预设阈值为10，那么可以判定这张图像上有4个虫点。In this embodiment, performing worm statistics on the classified RGB image data, preferably, counting a set of RGB image data with a wormhole, and calculating the number of wormholes and a preset threshold based on the number of wormholes obtained. The ratio, thereby determining the number of insect points based on the ratio. When the ratio is an integer, the integer value is used as the number of insects. When the ratio is a decimal, the decimal is rounded up, and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
在上述任一实施例中，优选地，预设阈值为10。In any of the above embodiments, preferably, the preset threshold is 10.
在该实施例中，预设阈值为10，说明10个RGB虫眼数据构成一个虫点。本领域技术人员应该理解，预设阈值为10但不限于此，由于不同作物的种植收到多种因素的影响，如地域、季节、土壤条件、气候等，因而产生的病虫害也会大有不同，从而经过实测统计后的预设阈值也会相应变化。In this embodiment, the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point. Those skilled in the art should understand that the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
在上述任一实施例中，优选地，无监督聚类算法为K均值聚类。In any of the above embodiments, preferably, the unsupervised clustering algorithm is K-means clustering.
在该实施例中，K均值聚类(K-means clustering)是最典型的聚类算法(当然，除此之外，还有很多诸如属于划分法K-MEDOIDS算法、CLARANS算法；属于层次法的BIRCH算法、CURE算法、CHAMELEON算法等；基于密度的方法：DBSCAN算法、OPTICS算法、DENCLUE算法等；基于网格的方法：STING算法、CLIQUE算法、WAVE-CLUSTER算法；基于模型的方法等)。本领域技术人员应该理解，在使用无监督聚类算法进行聚类分析时，也可以选择K均值聚类以外的其它无监督聚类算法。In this embodiment, K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.). Those skilled in the art should understand that when using the unsupervised clustering algorithm for cluster analysis, other unsupervised clustering algorithms other than K-means clustering can also be selected.
在上述任一实施例中，优选地，病虫害图像分类包括：虫眼数据组和非虫眼数据组。In any of the above embodiments, preferably, the pest image classification includes: a worm eye data set and a non-worm eye data set.
在该实施例中，通过无监督聚类算法对病虫害图像进行聚类分析后，得到虫眼数据组和非虫眼数据组，从而只需对虫眼数据组进行统计，以确认虫点的个数，实现病虫害的有效识别。In this embodiment, after the cluster analysis of the pest image by the unsupervised clustering algorithm, the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
如图4所示，根据本发明的一个实施例的病虫害识别装置的示意框图。 其中，该病虫害识别装置400包括：As shown in Figure 4, a schematic block diagram of a pest identification device in accordance with one embodiment of the present invention. The pest identification device 400 includes:
图像采集与处理单元402，用于采集病虫害图像，将病虫害图像转换成RGB图像数据；The image acquisition and processing unit 402 is configured to collect a pest image and convert the pest image into RGB image data.
病虫害识别单元404，用于使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；The pest and disease identification unit 404 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
统计单元406，用于对分类后的RGB图像数据进行虫眼统计，确定虫点数量；The statistic unit 406 is configured to perform worm statistics on the classified RGB image data to determine the number of insect spots;
判断单元408，用于判断虫点数量是否大于预设虫点数量；The determining unit 408 is configured to determine whether the number of insects is greater than a preset number of insects;
提醒单元410，用于当判断结果为是时，发出报警提示。The reminding unit 410 is configured to issue an alarm prompt when the determination result is yes.
在该实施例中，通过采集病虫害图像，优选的，通过摄像头拍摄粘虫板照片，并将照片解析成RGB格式(RGB格式为一种对颜色进行编码的方法，统称为“颜色空间”或“色域”)的像素数据，使用无监督聚类算法对这些像素数据进行聚类分析(聚类(Clustering)分析是无监督式机器学习(unsupervised learning)的一个典型应用，也是探索性数据挖掘中的一种常用方法，简单地说就是把相似的东西分到一组)，得到病虫害的分类，对分类后的数据进行虫眼统计，就能够确认这张图像上有多少虫点，当虫点的数量超过预设数值时，说明虫点较多，病虫害严重，发出报警提示。通过本发明的实施例，有效识别农业病虫害并及时的提醒相关人员进行病虫害防治，避免由于病虫害导致的农业减产、农产品质量下降、经济损失等问题。In this embodiment, by collecting the pest image, preferably, the photo of the sticky board is photographed by the camera, and the photo is parsed into an RGB format (the RGB format is a method of encoding the color, collectively referred to as "color space" or " Pixel data of the gamut", using unsupervised clustering algorithm to cluster these pixel data (clustering analysis is a typical application of unsupervised learning, and also in exploratory data mining A common method, which is simply to group similar things into groups, to get the classification of pests and diseases, and to perform statistics on the classified data, it is possible to confirm how many insect spots are on this image. When the number exceeds the preset value, it indicates that there are more insects, serious pests and diseases, and an alarm is issued. Through the embodiments of the present invention, agricultural pests and diseases are effectively identified, and relevant personnel are promptly reminded to carry out pest control, and problems such as agricultural production reduction, agricultural product quality degradation, and economic loss caused by pests and diseases are avoided.
如图5所示，根据本发明的一个实施例的病虫害识别装置的示意框图。其中，该病虫害识别装置500包括：As shown in Fig. 5, a schematic block diagram of a pest identification device according to an embodiment of the present invention. The pest identification device 500 includes:
图像采集与处理单元502，用于采集病虫害图像，将病虫害图像转换成RGB图像数据；The image acquisition and processing unit 502 is configured to collect a pest image and convert the pest image into RGB image data.
病虫害识别单元504，用于使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；The pest and disease identification unit 504 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
统计单元506，用于对分类后的RGB图像数据进行虫眼统计，确定虫点数量；The statistic unit 506 is configured to perform worm statistics on the classified RGB image data to determine the number of insect spots;
判断单元508，用于判断虫点数量是否大于预设虫点数量； The determining unit 508 is configured to determine whether the number of insects is greater than a preset number of insects;
提醒单元510，用于当判断结果为是时，发出报警提示；The reminding unit 510 is configured to issue an alarm prompt when the determination result is yes;
病虫害识别单元504具体包括：The pest identification unit 504 specifically includes:
建模单元5042，用于对RGB图像数据进行无监督聚类学习，建立相似度模型；分类单元5044，用于根据相似度模型，对RGB图像数据进行聚合分类。The modeling unit 5042 is configured to perform unsupervised clustering learning on the RGB image data to establish a similarity model, and the classification unit 5044 is configured to perform aggregation and classification on the RGB image data according to the similarity model.
在该实施例中，将病虫害图像解析为RGB图像数据后，通过对无监督聚类算法进行训练学习，从而建立最佳相似度模型，根据相似度模型，对RGB图像数据进行聚合分类，得到带虫眼的数据组以及不带虫眼的数据组，并在分类的基础上通过对带虫眼数据的统计分析，准确确认虫点个数，从而有效识别病虫害图像。In this embodiment, after the pest image is parsed into RGB image data, the unsupervised clustering algorithm is trained and learned to establish an optimal similarity model, and the RGB image data is classified and classified according to the similarity model to obtain a band. The data set of the insect eye and the data set without the insect eye, and the statistical analysis of the data of the insect eye on the basis of the classification, accurately confirm the number of insect spots, thereby effectively identifying the pest and disease image.
如图6所示，根据本发明的一个实施例的病虫害识别装置的示意框图。其中，该病虫害识别装置600包括：As shown in Fig. 6, a schematic block diagram of a pest identification device according to an embodiment of the present invention. The pest identification device 600 includes:
图像采集与处理单元602，用于采集病虫害图像，将病虫害图像转换成RGB图像数据；An image acquisition and processing unit 602 is configured to collect a pest image and convert the pest image into RGB image data;
病虫害识别单元604，用于使用无监督聚类算法对RGB图像数据进行聚类分析，得到病虫害图像分类；The pest and disease identification unit 604 is configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
统计单元606，用于对分类后的RGB图像数据进行虫眼统计，确定虫点数量；The statistic unit 606 is configured to perform worm statistics on the classified RGB image data to determine the number of insect points;
判断单元608，用于判断虫点数量是否大于预设虫点数量；The determining unit 608 is configured to determine whether the number of insect points is greater than a preset number of insect points;
提醒单元610，用于当判断结果为是时，发出报警提示；The reminding unit 610 is configured to issue an alarm prompt when the determination result is yes;
病虫害识别单元604具体包括：The pest and disease identification unit 604 specifically includes:
建模单元6042，用于对RGB图像数据进行无监督聚类学习，建立相似度模型；分类单元6044，用于根据相似度模型，对RGB图像数据进行聚合分类；The modeling unit 6042 is configured to perform unsupervised clustering learning on the RGB image data to establish a similarity model, and the classification unit 6044 is configured to perform aggregation and classification on the RGB image data according to the similarity model;
统计单元606具体包括：The statistics unit 606 specifically includes:
计数单元6062，用于统计虫眼的数量；计算单元6064，用于计算虫眼的数量与预设阈值的比值，根据比值确定虫点数量；计算单元6062，具体用于当比值为整数值时，将整数值作为虫点数量值；计算单元6064，具体还用于当比值为小数时，将小数进位取整，将取整后的整数值作为虫点数 量值。a counting unit 6062, configured to count the number of insect eyes; a calculating unit 6064, configured to calculate a ratio of the number of insect eyes to a preset threshold, and determine a number of insect points according to the ratio; and a calculating unit 6062, specifically, when the ratio is an integer value, The integer value is used as the number of bugs; the calculating unit 6064 is specifically used to round the decimals when the ratio is a decimal, and use the integer value as the number of insects. Measured value.
在该实施例中，对分类后的RGB图像数据的进行虫眼统计，优选的，对带有虫眼的一组RGB图像数据进行统计，在得到虫眼数量的基础上，计算虫眼数量与预设阈值的比值，从而根据比值确定虫点的数量。当比值为整数时，将该整数值作为虫点数，当比值为小数时，将小数进位取整，将取整后的数值作为虫点数。举例来说，统计虫眼点的个数为35，预设阈值为10，那么可以判定这张图像上有4个虫点。In this embodiment, performing worm statistics on the classified RGB image data, preferably, counting a set of RGB image data with a wormhole, and calculating the number of wormholes and a preset threshold based on the number of wormholes obtained. The ratio, thereby determining the number of insect points based on the ratio. When the ratio is an integer, the integer value is used as the number of insects. When the ratio is a decimal, the decimal is rounded up, and the rounded value is used as the number of insects. For example, if the number of statistical eye points is 35 and the preset threshold is 10, then it can be determined that there are 4 insect points on this image.
在上述任一实施例中，优选地，预设阈值为10。In any of the above embodiments, preferably, the preset threshold is 10.
在该实施例中，预设阈值为10，说明10个RGB虫眼数据构成一个虫点。本领域技术人员应该理解，预设阈值为10但不限于此，由于不同作物的种植收到多种因素的影响，如地域、季节、土壤条件、气候等，因而产生的病虫害也会大有不同，从而经过实测统计后的预设阈值也会相应变化。In this embodiment, the preset threshold is 10, indicating that 10 RGB worm eye data constitutes a pest point. Those skilled in the art should understand that the preset threshold is 10, but is not limited thereto. Since the cultivation of different crops is affected by various factors, such as region, season, soil condition, climate, etc., the pests and diseases will be greatly different. Therefore, the preset threshold after the measured statistics will also change accordingly.
在上述任一实施例中，优选地，无监督聚类算法为K均值聚类。In any of the above embodiments, preferably, the unsupervised clustering algorithm is K-means clustering.
在该实施例中，K均值聚类(K-means clustering)是最典型的聚类算法(当然，除此之外，还有很多诸如属于划分法K-MEDOIDS算法、CLARANS算法；属于层次法的BIRCH算法、CURE算法、CHAMELEON算法等；基于密度的方法：DBSCAN算法、OPTICS算法、DENCLUE算法等；基于网格的方法：STING算法、CLIQUE算法、WAVE-CLUSTER算法；基于模型的方法等)。本领域技术人员应该理解，在使用无监督聚类算法进行聚类分析时，也可以选择K均值聚类以外的其它无监督聚类算法。In this embodiment, K-means clustering is the most typical clustering algorithm (except, of course, there are many such as the K-MEDOIDS algorithm and the CLARANS algorithm belonging to the partitioning method; BIRCH algorithm, CURE algorithm, CHAMELEON algorithm, etc.; density-based methods: DBSCAN algorithm, OPTICS algorithm, DENCLUE algorithm, etc.; grid-based methods: STING algorithm, CLIQUE algorithm, WAVE-CLUSTER algorithm; model-based method, etc.). Those skilled in the art should understand that when using the unsupervised clustering algorithm for cluster analysis, other unsupervised clustering algorithms other than K-means clustering can also be selected.
在上述任一实施例中，优选地，病虫害图像分类包括：虫眼数据组和非虫眼数据组。In any of the above embodiments, preferably, the pest image classification includes: a worm eye data set and a non-worm eye data set.
在该实施例中，通过无监督聚类算法对病虫害图像进行聚类分析后，得到虫眼数据组和非虫眼数据组，从而只需对虫眼数据组进行统计，以确认虫点的个数，实现病虫害的有效识别。In this embodiment, after the cluster analysis of the pest image by the unsupervised clustering algorithm, the insect eye data group and the non-worm eye data group are obtained, so that only the insect eye data group is counted to confirm the number of insect spots, and the realization is realized. Effective identification of pests and diseases.
具体实施例，通过摄像头采集到蔬菜的照片，并把照片解析成RGB数据，对这些RGB数据进行无监督聚类(K-means)，这些RGB像素数据就会根据相似度聚合成一组一组的数据，然后找出相关虫点的那组数据进行统计，就能知道这张图有多少虫点。具体的步骤如下： In a specific embodiment, a photo of the vegetable is collected by the camera, and the photo is parsed into RGB data, and the RGB data is unsupervised clustered (K-means), and the RGB pixel data is aggregated into a group according to the similarity. The data, and then find out the set of data related to the insect point to count, you can know how many insect points in this picture. The specific steps are as follows:
首先，将采集到的照片转换一张图片为RGB值，其次，进行无监督聚类学习并建立相似度模型，RGB值经过聚类后会产生2种类别，一种是虫眼RGB数据，一种是非虫眼RGB数据，其中虫眼RGB数据类别为0，非虫眼RGB数据为1；最后，对虫眼数据进行统计，具体的为统计类别为0的个数为35，实测10像素点为一个虫点，经计算得知，这张图上有4个虫点。First, the captured photos are converted into a picture as RGB values. Secondly, unsupervised clustering learning is performed and a similarity model is established. After clustering, the RGB values are generated into two categories, one is the RGB data of the insect eye, and the other is It is non-worm eye RGB data, in which the RGB data category of the insect eye is 0, and the RGB data of the non-worm eye is 1; finally, the data of the insect eye is counted, specifically, the number of the statistical category is 0, and the measured 10 pixel is a pest point. It is calculated that there are 4 insect spots on this map.
在本说明书的描述中，术语“一个实施例”、“一些实施例”、“具体实施例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或特点包含于本发明的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不一定指的是相同的实施例或实例。而且，描述的具体特征、结构、材料或特点可以在任何的一个或多个实施例或示例中以合适的方式结合。In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "specific embodiments" and the like means that the specific features, structures, materials, or characteristics described in connection with the embodiments or examples are included in the present invention. At least one embodiment or example. In the present specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples.
以上所述仅为本发明的优选实施例而已，并不用于限制本发明，对于本领域的技术人员来说，本发明可以有各种更改和变化。凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。 The above description is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Claims (12)

  1. 一种病虫害识别方法，其特征在于，包括：A method for identifying pests and diseases, comprising:
    采集病虫害图像，将所述病虫害图像转换成RGB图像数据；Collecting pest and disease images, converting the pest image into RGB image data;
    使用无监督聚类算法对所述RGB图像数据进行聚类分析，得到病虫害图像分类；Clustering analysis of the RGB image data using an unsupervised clustering algorithm to obtain a pest and disease image classification;
    对分类后的所述RGB图像数据进行虫眼统计，确定虫点数量；Performing worm count on the classified RGB image data to determine the number of insect spots;
    判断所述虫点数量是否大于预设虫点数量；Determining whether the number of insect spots is greater than a preset number of insect spots;
    当判断结果为是时，发出报警提示。When the judgment result is yes, an alarm prompt is issued.
  2. 根据权利要求1所述的病虫害识别方法，其特征在于，所述使用无监督聚类算法对所述RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：The method for identifying pests and diseases according to claim 1, wherein the step of performing cluster analysis on the RGB image data using an unsupervised clustering algorithm to obtain a classification of pest and disease images comprises:
    对所述RGB图像数据进行无监督聚类学习，建立相似度模型；Performing unsupervised clustering learning on the RGB image data to establish a similarity model;
    根据所述相似度模型，对所述RGB图像数据进行聚合分类。The RGB image data is subjected to aggregation classification according to the similarity model.
  3. 根据权利要求1所述的病虫害识别方法，其特征在于，所述对分类后的所述RGB图像数据的进行虫眼统计，确定虫点数量的步骤，具体包括：The method for identifying pests and diseases according to claim 1, wherein the step of performing worm count on the classified RGB image data to determine the number of insect spots comprises:
    统计所述虫眼的数量；Counting the number of the insect eyes;
    计算所述虫眼的数量与预设阈值的比值，根据所述比值确定所述虫点数量；Calculating a ratio of the number of the insect eyes to a preset threshold, and determining the number of the insect points according to the ratio;
    当所述比值为整数值时，将所述整数值作为所述虫点数量值；When the ratio is an integer value, the integer value is used as the number of insect points;
    当所述比值为小数时，将所述小数进位取整，将取整后的整数值作为所述虫点数量值。When the ratio is a decimal, the decimal carry is rounded, and the rounded integer value is used as the insect number value.
  4. 根据权利要求3所述的病虫害识别方法，其特征在于，The pest and disease recognition method according to claim 3, characterized in that
    所述预设阈值为10。The preset threshold is 10.
  5. 根据权利要求1至4中任一项所述的病虫害识别方法，其特征在于，The pest and disease identification method according to any one of claims 1 to 4, characterized in that
    所述无监督聚类算法为K均值聚类。The unsupervised clustering algorithm is K-means clustering.
  6. 根据权利要求1至4中任一项所述的病虫害识别方法，其特征在于，The pest and disease identification method according to any one of claims 1 to 4, characterized in that
    所述病虫害图像分类包括：虫眼数据组和非虫眼数据组。The pest image classification includes: a worm eye data group and a non-worm eye data group.
  7. 一种病虫害识别装置，其特征在于，包括： A pest and disease identification device, comprising:
    图像采集与处理单元，用于采集病虫害图像，将所述病虫害图像转换成RGB图像数据；An image acquisition and processing unit, configured to collect a pest and disease image, and convert the pest image into RGB image data;
    病虫害识别单元，用于使用无监督聚类算法对所述RGB图像数据进行聚类分析，得到病虫害图像分类；a pest and disease identification unit, configured to perform cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a pest and disease image classification;
    统计单元，用于对分类后的所述RGB图像数据进行虫眼统计，确定虫点数量；a statistical unit, configured to perform worm statistics on the classified RGB image data to determine the number of insect points;
    判断单元，用于判断所述虫点数量是否大于预设虫点数量；a determining unit, configured to determine whether the number of the insect points is greater than a preset number of insect points;
    提醒单元，用于当判断结果为是时，发出报警提示。The reminding unit is configured to issue an alarm prompt when the judgment result is yes.
  8. 根据权利要求7所述的病虫害识别装置，其特征在于，所述病虫害识别单元使用无监督聚类算法对所述RGB图像数据进行聚类分析，得到病虫害图像分类的步骤，具体包括：The pest and disease identification device according to claim 7, wherein the pest and disease identification unit performs cluster analysis on the RGB image data by using an unsupervised clustering algorithm to obtain a step of classifying the pest image, specifically comprising:
    建模单元，用于对所述RGB图像数据进行无监督聚类学习，建立相似度模型；a modeling unit, configured to perform unsupervised clustering learning on the RGB image data, and establish a similarity model;
    分类单元，用于根据所述相似度模型，对所述RGB图像数据进行聚合分类。And a classification unit, configured to perform aggregation classification on the RGB image data according to the similarity model.
  9. 根据权利要求7所述的病虫害识别装置，其特征在于，所述统计单元对分类后的所述RGB图像数据的进行虫眼统计，确定虫点数量的步骤，具体包括：The device for identifying pests and diseases according to claim 7, wherein the step of the statistic unit performing worm count on the RGB image data after the classification, and determining the number of insect points, specifically includes:
    计数单元，用于统计所述虫眼的数量；a counting unit for counting the number of the insect eyes;
    计算单元，用于计算所述虫眼的数量与预设阈值的比值，根据所述比值确定所述虫点数量；a calculating unit, configured to calculate a ratio of the number of the insect eyes to a preset threshold, and determine the number of the insect points according to the ratio;
    所述计算单元，具体用于当所述比值为整数值时，将所述整数值作为所述虫点数量值；The calculating unit is specifically configured to use the integer value as the number of the insect point value when the ratio is an integer value;
    所述计算单元，具体还用于当所述比值为小数时，将所述小数进位取整，将取整后的整数值作为所述虫点数量值。The calculating unit is further configured to: when the ratio is a decimal, round the decimal carry, and use the rounded integer value as the number of the insects.
  10. 根据权利要求9所述的病虫害识别装置，其特征在于，The pest identification device according to claim 9, wherein
    所述预设阈值为10。The preset threshold is 10.
  11. 根据权利要求7至10中任一项所述的病虫害识别装置，其特征在于，所述无监督聚类算法为K均值聚类。 The pest and disease recognition apparatus according to any one of claims 7 to 10, wherein the unsupervised clustering algorithm is K-means clustering.
  12. 根据权利要求7至10中任一项所述的病虫害识别装置，其特征在于，所述病虫害图像分类包括：虫眼数据组和非虫眼数据组。 The pest identification device according to any one of claims 7 to 10, wherein the pest image classification comprises: a pest eye data set and a non-worm eye data set.
PCT/CN2017/086428 2016-12-30 2017-05-27 Method and apparatus for identifying disease and insect damage WO2018120634A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611259656.0A CN106650822A (en) 2016-12-30 2016-12-30 Identification method and device for diseases and insect pests
CN201611259656.0 2016-12-30

Publications (1)

Publication Number Publication Date
WO2018120634A1 true WO2018120634A1 (en) 2018-07-05

Family

ID=58837473

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/086428 WO2018120634A1 (en) 2016-12-30 2017-05-27 Method and apparatus for identifying disease and insect damage

Country Status (2)

Country Link
CN (1) CN106650822A (en)
WO (1) WO2018120634A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3654272A1 (en) * 2018-11-15 2020-05-20 Korea Institute of Science and Technology Crop injury diagnosis system and method
CN111985246A (en) * 2020-08-27 2020-11-24 武汉东湖大数据交易中心股份有限公司 Disease cognitive system based on main symptoms and accompanying symptom words
CN113505706A (en) * 2021-07-14 2021-10-15 陈小义 Automatic disease identification method and system based on AI intelligence
CN115285240A (en) * 2022-08-25 2022-11-04 大连海事大学 Agricultural plant protection trolley and control method

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650822A (en) * 2016-12-30 2017-05-10 深圳前海弘稼科技有限公司 Identification method and device for diseases and insect pests
CN112465038A (en) * 2020-11-30 2021-03-09 深圳市识农智能科技有限公司 Method and system for identifying disease and insect pest types of fruit trees

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130028487A1 (en) * 2010-03-13 2013-01-31 Carnegie Mellon University Computer vision and machine learning software for grading and sorting plants
CN102915446A (en) * 2012-09-20 2013-02-06 复旦大学 Plant disease and pest detection method based on SVM (support vector machine) learning
CN104881865A (en) * 2015-04-29 2015-09-02 北京林业大学 Forest disease and pest monitoring and early warning method and system based on unmanned plane image analysis
CN106650822A (en) * 2016-12-30 2017-05-10 深圳前海弘稼科技有限公司 Identification method and device for diseases and insect pests

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102214306B (en) * 2011-06-16 2013-01-30 中国农业大学 Leaf disease spot identification method and device
CN103034872A (en) * 2012-12-26 2013-04-10 四川农业大学 Farmland pest recognition method based on colors and fuzzy clustering algorithm
CN103390080B (en) * 2013-07-12 2016-06-08 北京农业信息技术研究中心 A kind of plant disease speckle color simulation method and device
CN103489006A (en) * 2013-10-11 2014-01-01 河南城建学院 Computer vision-based rice disease, pest and weed diagnostic method
CN104598908B (en) * 2014-09-26 2017-11-28 浙江理工大学 A kind of crops leaf diseases recognition methods
CN105787446A (en) * 2016-02-24 2016-07-20 上海劲牛信息技术有限公司 Smart agricultural insect disease remote automatic diagnosis system

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130028487A1 (en) * 2010-03-13 2013-01-31 Carnegie Mellon University Computer vision and machine learning software for grading and sorting plants
CN102915446A (en) * 2012-09-20 2013-02-06 复旦大学 Plant disease and pest detection method based on SVM (support vector machine) learning
CN104881865A (en) * 2015-04-29 2015-09-02 北京林业大学 Forest disease and pest monitoring and early warning method and system based on unmanned plane image analysis
CN106650822A (en) * 2016-12-30 2017-05-10 深圳前海弘稼科技有限公司 Identification method and device for diseases and insect pests

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3654272A1 (en) * 2018-11-15 2020-05-20 Korea Institute of Science and Technology Crop injury diagnosis system and method
CN111985246A (en) * 2020-08-27 2020-11-24 武汉东湖大数据交易中心股份有限公司 Disease cognitive system based on main symptoms and accompanying symptom words
CN111985246B (en) * 2020-08-27 2023-08-15 武汉东湖大数据交易中心股份有限公司 Disease cognitive system based on main symptoms and accompanying symptom words
CN113505706A (en) * 2021-07-14 2021-10-15 陈小义 Automatic disease identification method and system based on AI intelligence
CN115285240A (en) * 2022-08-25 2022-11-04 大连海事大学 Agricultural plant protection trolley and control method

Also Published As

Publication number Publication date
CN106650822A (en) 2017-05-10

Similar Documents

Publication Publication Date Title
WO2018120634A1 (en) Method and apparatus for identifying disease and insect damage
Aquino et al. A new methodology for estimating the grapevine-berry number per cluster using image analysis
Liu et al. A computer vision system for early stage grape yield estimation based on shoot detection
Sarangdhar et al. Machine learning regression technique for cotton leaf disease detection and controlling using IoT
CN105938564B (en) Rice disease identification method and system based on principal component analysis and neural network
JP2019125340A (en) Systems and methods for automated inferencing of changes in spatiotemporal images
RU2018143340A (en) RECOGNITION OF WEED IN THE NATURAL ENVIRONMENT
Gulhane et al. Diagnosis of diseases on cotton leaves using principal component analysis classifier
BR112018009108B1 (en) METHOD FOR ACQUISITION AND ANALYSIS OF AERIAL IMAGES
Renugambal et al. Application of image processing techniques in plant disease recognition
CN103489006A (en) Computer vision-based rice disease, pest and weed diagnostic method
DE112009000480T5 (en) Dynamic object classification
SE1930281A1 (en) Method for calculating deviation relations of a population
CN112257702A (en) Crop disease identification method based on incremental learning
Ji et al. In-field automatic detection of maize tassels using computer vision
Pinto et al. Crop disease classification using texture analysis
CN107622236A (en) Based on bee colony and gradient lifting decision Tree algorithms crops disease diagnosing method for early warning
Ye et al. An image-based approach for automatic detecting tasseling stage of maize using spatio-temporal saliency
Ramesh et al. Detection of rows in agricultural crop images acquired by remote sensing from a UAV
Kumar et al. An identification of crop disease using image segmentation
CN115601670A (en) Pine wilt disease monitoring method based on artificial intelligence and high-resolution remote sensing image
US11150657B2 (en) Lossy data compressor for vehicle control systems
Saini et al. Multiclass Classification of Rice Leaf Disease Using Deep Learning Based Model
CN110796148B (en) Litchi insect pest monitoring and identifying system and litchi insect pest monitoring and identifying method
Hussain et al. Classification and detection of plant disease using feature extraction methods

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17887086

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS (EPO FORM 1205A DATED 25.10.2019)

122 Ep: pct application non-entry in european phase

Ref document number: 17887086

Country of ref document: EP

Kind code of ref document: A1