CN114581792A

CN114581792A - Agricultural disaster monitoring method and system based on satellite remote sensing image

Info

Publication number: CN114581792A
Application number: CN202210221911.1A
Authority: CN
Inventors: 李庆浩
Original assignee: Shandong Origin Information Technology Co ltd
Current assignee: Shandong Origin Information Technology Co ltd
Priority date: 2022-03-09
Filing date: 2022-03-09
Publication date: 2022-06-03

Abstract

The invention relates to a method for monitoring agricultural disasters based on satellite remote sensing images, which comprises the following steps: (1) a first image is obtained, wherein each pixel point of the first image is an n-dimensional vector [ f1, f2, f3,. talka., fn ], and each value in the vector corresponds to a band1, a band2, a band3, … and a band; (2) determining the number of categories, namely determining the number h of categories to be generated; (3) a first model training step, wherein each pixel point in a first image is input into a Gaussian mixture model GMM to be trained to obtain a second image, and each pixel of the second image is assigned with a pixel value, wherein the pixel value is in a category from 1 to h; the method has the advantages that places suspected of diseases and insect pests in the satellite images can be found quickly, agricultural personnel can mark the places with the diseases and insect pests on the images quickly only by simple selection, workload of people is reduced to a great extent, omission is not prone to occurring, and a technical basis is provided for accurate management of agriculture.

Description

Agricultural disaster monitoring method and system based on satellite remote sensing image

Technical Field

The invention relates to a crop monitoring method, in particular to an agricultural disaster monitoring method, device and equipment based on satellite remote sensing images and a computer readable storage medium.

Background

The description of the background of the invention pertaining to the present invention is intended only for the purpose of illustration and for the purpose of facilitating an understanding of the summary of the invention, and should not be taken as an admission or admission that the applicant is aware of or is aware of the prior art at the date of filing this application as first filed.

The technology of aviation, aerospace and satellite remote sensing is continuously developed since the last century, and great convenience is provided for agricultural planting and management through satellite remote sensing image data. With the development of the satellite remote sensing field, a certain history of agricultural management is provided through satellite images, but the satellite images generally have large coverage area, are simply checked manually, have huge marking workload and are easy to omit.

Disclosure of Invention

The present invention is directed to solving, at least to some extent, one of the problems in the related art.

In order to overcome the defects in the prior art, embodiments of the present invention provide an agricultural disaster monitoring method based on satellite remote sensing images, a crop classification device, a computer device, and a computer-readable storage medium.

Therefore, the agricultural disaster monitoring method based on the satellite remote sensing image comprises the following steps: (1) a first image is obtained, wherein each pixel point of the first image is an n-dimensional vector [ f1, f2, f3,. talka., fn ], and each value in the vector corresponds to a band1, a band2, a band3, … and a band; (2) determining the number of categories, namely determining the number h of categories to be generated; (3) a model training step, namely inputting each pixel point in the first image into a Gaussian Mixture Model (GMM) for training to obtain a second image, and distributing a pixel value to each pixel of the second image, wherein the pixel value is a category from 1 to h; and (4) outputting the classification result, and outputting the second image.

According to a preferred embodiment, among others, it further comprises: (5) determining a reclassification category, namely analyzing h categories of the second image and determining a reclassification specific category in the h categories; (6) determining the number of reclassifications, namely determining the number k of categories to be generated by the specific categories of the reclassification; (7) a model training step, in which all pixel points of partial categories in a second image are input into a Gaussian mixture model GMM to be trained to obtain a third image, and each pixel of the third image is assigned with a pixel value, wherein the pixel value is a category from 1 to k; (8) a step of outputting a classification result, namely outputting a third image; and (9) iterating steps (5) through (8) until convergence; (10) and (4) an output result image unit, which combines all the pixels in the target specific category obtained in the step (5) each time to obtain an output final monitoring result image.

According to a preferred embodiment, among others, it further comprises: a characteristic engineering step, namely utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny wave bands are selected from the new feature to be combined and calculated mutually.

According to a preferred embodiment, wherein the feature engineering step specifically comprises: utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny two wave bands are selected to be mutually combined and calculated to generate h (h-1)/2 new characteristics.

According to a preferred embodiment, wherein the feature engineering step specifically comprises: and selecting a specific small window to slide gradually on the original image, calculating one or more of the mean value, standard deviation, value range, entropy, mutual information and the like of each wave band in the range of the sliding window, and generating a new characteristic as a new characteristic of the central position of the sliding window.

To this end, the agricultural disaster monitoring device based on the satellite remote sensing image according to one embodiment of the present invention comprises: a first image obtaining unit, configured to obtain a first image, where a value of each pixel point of the first image is an n-dimensional vector [ f ]₁,f₂,f₃，...，f_n]Each value in the vector corresponds to a band₁，band₂，band₃，…，band_n(ii) a A unit for determining the number of classes to be generatedThe number of categories h; the model training unit is used for inputting each pixel point in the first image into a Gaussian Mixture Model (GMM) for training to obtain a second image, and distributing a pixel value to each pixel of the second image, wherein the pixel value is in a category from 1 to h; and an output classification result unit for outputting the second image.

According to a preferred embodiment, among others, it further comprises: (5) a classification determining unit which analyzes the h classes of the second image and determines the specific classification of the h classes; (6) the device comprises a unit for determining the number of reclassifications, a unit for determining the number k of categories to be generated by specific categories of the reclassification; (7) the model training unit is used for inputting all pixel points of partial categories in the second image into a Gaussian Mixture Model (GMM) for training to obtain a third image, and distributing a pixel value to each pixel of the third image, wherein the pixel value is a category from 1 to k; (8) a classification result output unit which outputs a third image; and (9) iterating the units (5) to (8) until convergence; and (10) an output result image unit, which combines all the pixels in the target specific category obtained by the unit (5) each time to obtain an output final monitoring result image.

According to a preferred embodiment, among others, it further comprises: a characteristic engineering unit for utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny wave bands are selected from the new feature to be combined and calculated mutually.

According to a preferred embodiment, wherein the feature engineering unit is specifically: utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny two wave bands are selected to be mutually combined and calculated to generate h (h-1)/2 new characteristics.

According to a preferred embodiment, wherein the feature engineering unit is specifically: selecting a specific small window to slide gradually on the original image, calculating one or more of the mean value, standard deviation, value range, entropy, mutual information and the like of each wave band in the range of the sliding window, and generating a new feature which is used as a new feature of the center position of the sliding window.

Furthermore, the present invention provides a computer-readable storage medium, on which a computer program is stored, characterized in that the program, when executed by a processor, implements the steps of the above method.

Furthermore, the present invention provides a computer device comprising a memory, a processor and a computer program stored on the memory and running on the processor, wherein the processor implements the steps of the above method when executing the program.

The invention develops a set of algorithm based on a Gaussian Mixture Model (GMM) in a machine learning algorithm, and the algorithm can be used for analyzing pixels in remote sensing images to detect the crop condition.

For example, the method can quickly find the suspected disease and insect pest places in the satellite images, agricultural personnel can quickly mark the places with the disease and insect pests on the images only by simple selection, so that the workload of the personnel is reduced to a great extent, omission is not easy to occur, and a technical basis is provided for accurate agricultural management.

Preferably, the resolution of the satellite imagery is 1m-30 m. According to the observation that a single pixel of the satellite image under the resolution can be regarded as sampling of Gaussian distribution with different mean values and variances, the method adopts the Gaussian mixture model to perform agricultural pest clustering on the satellite image.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

In order to more clearly illustrate the technical solution of the present invention, the drawings used in the description of the embodiments will be briefly described as follows:

fig. 1 shows a flow chart of an agricultural disaster monitoring method based on satellite remote sensing images according to the present application;

FIG. 2 shows a block diagram of an agricultural disaster monitoring device based on satellite remote sensing images of the present application;

FIG. 3 shows a flow chart of another method for monitoring agricultural disasters based on satellite remote sensing images according to the application;

FIG. 4 shows a block diagram of another satellite remote sensing image based agricultural disaster monitoring device of the present application;

FIG. 5 shows a flow chart of yet another method for agricultural disaster monitoring based on satellite remote sensing images of the present application;

FIG. 6 shows a block diagram of yet another satellite remote sensing image based agricultural disaster monitoring device of the present application;

FIG. 7 illustrates a block diagram of an exemplary computer device suitable for use to implement embodiments of the present application; and

fig. 8(a) - (b) sequentially show an original satellite image and a crop monitoring result schematic diagram generated by the agricultural disaster monitoring method or device based on the satellite remote sensing image.

The foregoing and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout.

The agricultural disaster monitoring method and device based on the satellite remote sensing image in the embodiment of the invention are described below with reference to the accompanying drawings.

ExamplesA

Reference is made to figure 1. In this embodiment, the agricultural disaster monitoring method based on the satellite remote sensing image includes:

(1) a step of acquiring a first image, each image of said first imageTaking the value of a prime point as a vector [ f ] of dimension n₁,f₂,f₃，...，f_n]Each value in the vector corresponds to a band₁，band₂，band₃，…，band_n。

First, a satellite image of an area to be monitored is acquired. The method can directly operate the satellite image without other special pretreatment.

For example, satellite image data, i.e., a first image, captured by a satellite is obtained, and each pixel in the satellite image takes a 7-dimensional vector [ f [ ]₁,f₂,f₃，...,f₇]Then each pixel in the satellite image can be represented as pix [ f ]₁,f₂,f₃，...，f₇]Each value in the vector corresponds to a band.

(2) And determining the number of the classes, namely determining the number h of the classes to be generated.

The number of classes to be generated, e.g. h classes, is selected and then all pixel samples are used as input for the GMM model in the subsequent step. It should be noted that h merely indicates the number of categories, and the specific meaning of the categories does not need to be defined.

(3) And a model training step, namely inputting each pixel point in the first image into a Gaussian Mixture Model (GMM) for training to obtain a second image, and distributing a pixel value to each pixel point of the second image, wherein the pixel value is in a category from 1 to h.

Specifically, each pixel point is a sample, and each sample is a multidimensional vector [ f [ ]₁,f₂,f₃，...，f₇]And (3) each value in the multidimensional vector corresponds to one band described in the step (2), and the number of the categories (such as h categories) generated in the step (2) is used as GMM model input by all pixel points in the first image.

The Gaussian Mixture Model (GMM) is an unsupervised clustering learning algorithm, which accurately quantifies objects with gaussian probability density functions (normal distribution curves) and decomposes one object into a plurality of models formed based on the gaussian probability density functions (normal distribution curves). Colloquially, regardless of how the observed data set is distributed and how it exhibits, it can be fitted by a mixture of multiple single gaussian models.

The agricultural disaster monitoring method based on the satellite remote sensing image is formed by combining GMM, and a good classification effect can be achieved without manually marking a training sample. The contents related to the algorithm will be described below.

1. Multidimensional gaussian (normal) distribution

A normal distribution is a distribution quite common in nature, and a multidimensional normal distribution is determined by a mean vector 11 μ and a covariance matrix Σ. A normal distribution is disclosed as follows:

in the above formula, x is a column vector with dimension d, u is a model expectation, and Σ is a model variance. In practical applications u is usually replaced by the mean value of the samples, and Σ is usually replaced by the difference in the samples. It is easy to determine whether a sample x belongs to class C. Since each class has its own u and Σ, substituting x into the formula, we consider x to belong to class C when the probability is greater than a certain threshold.

2. Gaussian mixture model

Different mu and sigma can obtain different gaussian distributions, the gaussian mixture model requires the distribution p (x) of a random variable x, and if a sample (corresponding to each pixel point in the patent) is to be finally clustered into n classes, one gaussian distribution p (x | c ═ i) corresponding to each class is represented by the i-th class. The following can be obtained by using a total probability formula:

the Gaussian mixture model is based on a sample set X ═ X_i|i∈[1，m]Solving p (c ═ i) and N (x; μ i, Σ i) by the expectation-maximization algorithm. This results in a distribution of random variables x. Thus for any one xp. The probability of belonging to any class can be found:

p(c＝i|xp)∝p(xp|c＝i)p(c＝i)

that is, the probability that xp belongs to each class is obtained, and xp may be classified as a class corresponding to the corresponding maximum probability.

Therefore, by applying the GMM model in the step, all the pixel points can be divided into h categories according to the predetermined category number and the clustering result.

(4) And outputting a classification result, namely outputting a second image.

And finally, the output result of the GMM model is a picture with the same size as the input image, and the pixel value 1-h corresponds to the classification of each pixel point in the original image by the algorithm.

The model designed by the invention aims to accelerate the application of satellite images in monitoring, such as agricultural pest and disease damage, and simultaneously solves a great amount of heavy marking work of manpower, so that the prevention and treatment period of the agricultural pest and disease damage becomes rapid, and the management range is wider.

Referring to fig. 2, another example of the embodiment is an agricultural disaster monitoring device based on satellite remote sensing images, which specifically includes:

a first image obtaining unit, configured to obtain a first image, where a value of each pixel point of the first image is an n-dimensional vector [ f ]₁,f₂,f₃，...，f_n]Each value in the vector corresponds to a band₁，band₂，band₃，…，band_n(ii) a A classification number determining unit for determining the number h of classes to be generated; the model training unit is used for inputting each pixel point in the first image into a Gaussian Mixture Model (GMM) for training to obtain a second image, and distributing a pixel value to each pixel of the second image, wherein the pixel value is in a category from 1 to h; and an output classification result unit for outputting the second image.

It should be noted that the foregoing explanation of the method embodiment is also applicable to the apparatus of this embodiment, and details thereof are not repeated here.

The first embodiment can be used for analyzing pixels in the remote sensing image to detect the crop condition. For example, the method can quickly find the suspected pest and disease damage places in the satellite images, agricultural personnel can quickly mark the places with the pest and disease damage on the images only by simply selecting the places from the second images output after GMM training, so that the workload of the personnel is reduced to a great extent, omission is not easy to occur, and a technical basis is provided for accurate agricultural management.

ExamplesII

The embodiment considers that if the classification result of the GMM is not ideal enough, the scheme in the embodiment can be adopted to further optimize the monitoring result.

For example, 30 categories are selected, the GMM analyzes the output image after giving the second image, and two categories 1-4 and 1-6 are selected to meet the requirements. While continuing to analyze the second image, it is found that there are some actually needed results in the three categories 1-5, 1-7, 1-10, but the three categories 1-5, 1-7, 1-10 contain not only the desired results but also some other unrelated pixels.

As shown in fig. 3, the method further comprises:

(5) and determining a reclassification category, namely analyzing the h categories of the second image, and determining a target specific category and a reclassified specific category in the h categories.

For example, the number of categories to be classified is determined, for example, 10 categories are selected, i.e., h is 10.

(6) And determining the number of reclassifications, namely determining the number k of categories to be generated by the specific categories of the reclassification.

And analyzing whether the target pixel points and the non-target pixel points are clearly divided in all the categories obtained by the third image. For example, only categories 1-5, 1-7, 1-10 are determined, i.e., k is 3. The three classes include target pixel points and also include a plurality of irrelevant pixel points. And the other 25 types of data except the types 1-4, 1-6, 1-5, 1-7 and 1-10 are discarded because the data are irrelevant data, and the subsequent steps are not needed any more.

(7) And a model training step, namely inputting all pixel points of partial categories in the second image into a Gaussian mixture model GMM for training to obtain a third image, and distributing a pixel value for each pixel of the third image, wherein the pixel value is the category from 1 to k.

Then only corresponding pixels belonging to the three classes 1-5, 1-7, 1-10 are input into the GMM model for retraining.

(8) And outputting a classification result, namely outputting a third image.

The resultant, third image is then output. In this example, 10 classes of pixels, such as 3-1, 3-2, 3-3, …, and 3-10, are obtained after GMM training of pixels corresponding to the three classes 1-5, 1-7, and 1-10.

(9) And (5) iterating, and repeating the steps (5) to (8) until convergence.

And continuously analyzing all the categories obtained by the third image to determine whether the target pixel points and the non-target pixel points are clearly divided. If the partial category segmentation is still unclear, repeating the steps (5) to (8), and iterating and analogizing until convergence.

For example, in the 3 rd-1, 3 rd-2, 3 rd-3, …, 3 rd-10 categories, the 3 rd-2 and3 rd-3 categories are target pixels, and the 3 rd-5 and3 th-6 categories are indistinct categories, that is, the target pixels are included, and many irrelevant pixels are also included. Then iterative training continues for classes 3-5, 3-6, while the remaining completely irrelevant classes are discarded. And iterating and analogizing until convergence.

(10) And (4) a target pixel merging step, namely merging all the pixels in the target specific category obtained in the step (5) every time to obtain a target pixel set. .

In the above example, the 1 st-4 th, 1 st-6 th, 3 rd-2 th, 3 rd-3 rd, … … th images are all merged to obtain the output final monitoring result image.

The method aims to identify the possible problem places in the satellite images semi-automatically by utilizing computing resources through a machine learning method, and meanwhile, the problem parts in the satellite images are quickly identified through classification result analysis, and for the places with unclear classification, the algorithm can be executed iteratively to achieve accurate classification. Meanwhile, the work of professional personnel is greatly reduced and simplified.

Referring to fig. 4, another example of the embodiment is an agricultural disaster monitoring device based on satellite remote sensing images, and specifically, the device further includes, on the basis of the example of the agricultural disaster monitoring device based on satellite remote sensing images described in embodiment 1: a classification determining unit which analyzes the h classes of the second image and determines the specific classification of the h classes; the device comprises a unit for determining the number of reclassifications, a unit for determining the number k of categories to be generated by specific categories of the reclassification; the model training unit is used for inputting all pixel points of partial categories in the second image into a Gaussian Mixture Model (GMM) for training to obtain a third image, and distributing a pixel value to each pixel of the third image, wherein the pixel value is a category from 1 to k; a classification result output unit which outputs a third image; an iteration unit that repeats the units (5) to (8) until convergence; and (10) a target pixel merging unit for merging all the pixels in the target specific category obtained by the unit (5) each time to obtain a target pixel set.

Examples3

In order to further improve the classification accuracy, the invention can add related steps or units of feature engineering.

As shown in fig. 5, the method further includes: a characteristic engineering step, namely utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny wave bands are selected from the new feature to be combined and calculated mutually.

Preferably, the feature engineering step may be to use a band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny two wave bands are selected to be mutually combined and calculated to generate h (h-1)/2 new features.

For example, a step of selecting any two bands to be mutually combined to calculate and generate a new feature, and a step of mutually combining the band values of each pixel point to generate a new feature, for example, selecting any two bands from the selected band to be mutually combined to calculate and generate a new feature, thereby generating h (h-1)/2 new features. For example, two bands p1 in the first image are arbitrarily selected, and p2 is divided to generate new features, i.e., p1/p2(p2/p1 is also possible, but only one of the two is adopted), so that if the second image has n bands, n (n-1)/2 new features can be generated. Through the characteristic engineering, more new characteristic modes are generated by combining with each other, the number of the characteristics is reduced, and the operation speed is increased.

Preferably, the feature engineering step may be to select a specific small window to gradually slide on the original image, and calculate one or more of a mean value, a standard deviation, a value range, entropy, mutual information, and the like of each band in the range of the sliding window to generate a new feature, which is used as a new feature of the center position of the sliding window.

For example, a specific small window is selected to slide gradually on the original image, and one or more of the mean value, the standard deviation, the value range, the entropy, the mutual information and the like of each wave band are calculated in the range of the sliding window, so as to generate a new feature, which is used as a sample of the center position of the sliding window. For example, a specific window size (e.g., 5 × 5) is selected to slide on the original image, and the mean value, standard deviation, value range (maximum value minus minimum value), entropy, mutual information, and the like (not all generated, some of them may be selected, e.g., the first three generated) are calculated for each band within the sliding window to generate new features. For example, the second image has n bands, and each band calculates three new features including a mean value, a standard deviation, and a value range, so that the n bands generate 3n new features. At this time, the sliding window center point is added to all k wave bands corresponding to the original image, and a total of k + n (n-1)/2+3n features are used as a sample of the sliding window center position. Further, in the step of generating new features by the sliding window, the sliding window is used for extracting the spatial features, and a convolution algorithm in a sklern packet can be adopted for carrying out accelerated operation.

According to the preferred embodiment, the mode that the small window is used for taking the pixel feature points around the small range to generate more new features of the central point is adopted, noise is filtered, the features of adjacent pixel points are collected, the wave band features which are arranged at the front are synchronously depicted at multiple angles, so that the pixel points are described more clearly, and the generated picture is more real.

Fig. 6 shows a block diagram of an agricultural disaster monitoring device based on satellite remote sensing images in embodiment 3.

As shown in fig. 6, the apparatus may further include, on the basis of embodiment 1: a feature engineering unit for utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny wave bands are selected from the new feature to be combined and calculated mutually.

The specific implementation manner of the unit refers to the manner of combining, calculating and generating any two wave bands or the manner of generating a sliding window.

FIG. 7 illustrates a block diagram of an exemplary computer device suitable for use to implement embodiments of the present application. The computer device 12 shown in fig. 7 is only an example, and should not bring any limitation to the function and the scope of use of the embodiments of the present application.

As shown in FIG. 7, computer device 12 is in the form of a general purpose computing device. The components of computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.

Bus 18 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. These architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus, to name a few.

Computer device 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by computer device 12 and includes both volatile and nonvolatile media, removable and non-removable media.

Memory 28 may include computer system readable media in the form of volatile Memory, such as Random Access Memory (RAM) 30 and/or cache Memory 32. Computer device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 3, and commonly referred to as a "hard drive"). Although not shown in FIG. 7, a magnetic disk drive for reading from and writing to a removable nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable nonvolatile optical disk (e.g., a Compact disk Read Only Memory (CD-ROM), a Digital versatile disk Read Only Memory (DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the application.

A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally perform the functions and/or methodologies of the embodiments described herein.

Computer device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), with one or more devices that enable a user to interact with computer device 12, and/or with any devices (e.g., network card, modem, etc.) that enable computer device 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface 22. Moreover, computer device 12 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public Network such as the Internet) via Network adapter 20. As shown, network adapter 20 communicates with the other modules of computer device 12 via bus 18. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with computer device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

The processing unit 16 executes various functional applications and data processing, for example, implementing the methods mentioned in the foregoing embodiments, by executing programs stored in the system memory 28.

Referring to fig. 8, fig. 8(a) shows an original satellite image, and fig. 8(b) shows a schematic view of a monitoring analysis result generated by the remote sensing satellite image monitoring method of the present application.

In order to implement the foregoing embodiments, the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the state prediction method according to the foregoing method embodiments is implemented.

In order to implement the above-mentioned embodiments, the invention also proposes a non-transitory computer-readable storage medium on which a computer program is stored which, when executed by a processor, implements the state prediction method as described in the aforementioned method embodiments.

In order to implement the above embodiments, the present invention further proposes a computer program product, wherein when the instructions of the computer program product are executed by a processor, the state prediction method according to the foregoing method embodiments is implemented.

While various embodiments of the invention have been described above, it will be understood by those skilled in the art that various embodiments may be substituted or combined, and the invention is therefore intended to include all possible combinations of the same and/or different embodiments described. Thus, if one embodiment includes feature A, B, C and another embodiment includes feature B, D, then the invention should also be construed as including embodiments that include A, B, C, D in all other possible combinations, even though such embodiments may not be explicitly recited in the following text.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.

Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.

The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fibre device, a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, and a magneto-optical disk, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic or optical card, a nano-system (including molecular memory ICs), or any type of media or device suitable for storing instructions and/or data.

Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.

It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.

It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.

In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a separate product, may also be stored in a computer-readable storage medium. The "module" and "unit" in this specification refer to software and/or hardware that can perform a specific function independently or in cooperation with other components, where the hardware may be, for example, an FPGA (Field-Programmable gate array), an IC (Integrated Circuit), or the like.

The foregoing description is only exemplary of the preferred embodiments of the invention and is not intended to limit the invention in any way as to its nature or form. Although the present invention has been described with reference to the preferred embodiments, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention. However, any simple modification, equivalent replacement, improvement and the like of the above embodiments according to the technical spirit of the present invention should be included in the protection scope of the present invention without departing from the spirit and principle of the present invention.

Claims

1. The method for monitoring the agricultural disaster based on the satellite remote sensing image comprises the following steps:

(1) a step of obtaining a first image, which is to obtain the first image, wherein each pixel point of the first image is an n-dimensional vector [ f₁,f₂,f₃，...,f_n]Each value in the vector corresponds to a band₁，band₂，band₃，…，band_n；

(2) Determining the number of categories, namely determining the number h of categories to be generated;

(3) a first model training step, wherein each pixel point in a first image is input into a Gaussian mixture model GMM to be trained to obtain a second image, and each pixel of the second image is assigned with a pixel value, wherein the pixel value is in a category from 1 to h; and

(4) and a first classification result output step, namely outputting a second image.

2. The method for monitoring the agricultural disaster based on the satellite remote sensing image as claimed in claim 1, further comprising:

(5) determining a reclassification category, namely analyzing h categories of the second image and determining a specific reclassification category in the h categories;

(6) determining the number of reclassifications, namely determining the number k of categories to be generated by the specific categories of the reclassification;

(7) a second model training step, wherein all pixel points of partial categories in a second image are input into a Gaussian mixture model GMM to be trained to obtain a third image, and each pixel of the third image is assigned with a pixel value, wherein the pixel value is a category from 1 to k;

(8) a second step of outputting a classification result, namely outputting a third image;

(9) an iteration step, repeating the steps (5) to (8) until convergence; and

(10) and (5) a target pixel merging step, namely merging all the pixels in the target specific category obtained in the step (5) every time to obtain a target pixel set.

3. The method for monitoring agricultural disasters based on the satellite remote sensing images according to claim 1 or 2, characterized by further comprising:

a characteristic engineering step, namely utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny wave bands are selected from the new feature to be combined and calculated mutually.

4. The agricultural disaster monitoring method based on the satellite remote sensing images as claimed in claim 3, wherein the characteristic engineering steps specifically comprise:

utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nAny two wave bands are selected to be mutually combined and calculated to generate h (h-1)/2 new characteristics.

5. The agricultural disaster monitoring method based on the satellite remote sensing images as claimed in claim 3, wherein the characteristic engineering steps specifically comprise:

selecting a specific small window to slide gradually on the original image, calculating one or more of the mean value, standard deviation, value range, entropy, mutual information and the like of each wave band in the range of the sliding window, and generating a new feature which is used as a new feature of the center position of the sliding window.

6. An agricultural disaster monitoring device based on satellite remote sensing images comprises: a first image obtaining unit, configured to obtain a first image, where a value of each pixel point of the first image is an n-dimensional vector [ f ]₁,f₂,f₃，...,f_n]Each value in the vector corresponds to a band₁，band₂，band₃，…，band_n；

A classification number determining unit for determining the number h of classes to be generated;

the first model training unit is used for inputting each pixel point in the first image into a Gaussian Mixture Model (GMM) for training to obtain a second image, and distributing a pixel value to each pixel of the second image, wherein the pixel value is a category from 1 to h; and

and the first output classification result unit is used for outputting the second image.

7. The agricultural disaster monitoring device based on satellite remote sensing images as claimed in claim 6, further comprising:

(5) a classification determining unit which analyzes the h classes of the second image and determines the specific classification of the h classes;

(6) a determining reclassification number unit which determines the number k of categories to be generated by the specific categories of reclassification;

(7) the second model training unit is used for inputting all pixel points of partial categories in the second image into a Gaussian Mixture Model (GMM) for training to obtain a third image, and distributing a pixel value to each pixel of the third image, wherein the pixel value is a category from 1 to k;

(8) a second output classification result unit which outputs a third image;

(9) an iteration unit for repeatedly executing the units (5) to (8) until convergence; and

(10) and (5) a target pixel merging unit, merging all the pixels in the target specific category obtained in the step (5) each time to obtain a target pixel set.

8. An agricultural disaster monitoring device based on satellite remote sensing images as claimed in claim 6 or 7, further comprising:

a characteristic engineering unit for utilizing the band value band of each pixel point in the first image₁，band₂，band₃，…，band_nSelect arbitrary waveThe segments are computed in combination with each other to generate new features.

9. The agricultural disaster monitoring device based on the satellite remote sensing image according to claim 8, wherein the characteristic engineering unit is specifically:

10. The agricultural disaster monitoring device based on the satellite remote sensing image according to claim 8, wherein the characteristic engineering unit is specifically:

11. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1-5 are implemented when the program is executed by the processor.

12. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 5.