A method for extracting waterlogged corn based on satellite remote sensing

Abstract

The present application belongs to the technical field of agricultural disaster monitoring, and specifically relates to a method for extracting flood-damaged corn based on satellite remote sensing, including the following steps: obtaining field marker data in the target area, optical remote sensing image data that is not affected by floods and after being eroded by floods; extracting spatial distribution data of corn planting; establishing identification marks of flood-damaged corn and non-flood-damaged corn on the image; obtaining the spectral curves of flood-damaged corn and non-flood-damaged corn, and the change characteristics of the spectral curves of flood-damaged corn over time; obtaining the common spectral features that distinguish flood-damaged corn from non-flood-damaged corn; constructing a relationship for extracting flood-damaged corn index; and obtaining a binary distribution map of flood-damaged corn. The present application is a method for extracting flood-damaged corn based on the common features of flood-damaged corn remote sensing images and the change characteristics of spectral curves. It fully considers the complexity and diversity of flood-damaged corn image features, making the extraction of flood-damaged areas more comprehensive and the accuracy of flood remote sensing identification higher.

Description

Technical Field

[0001] The present application belongs to the technical field of agricultural disaster monitoring, and specifically relates to a method for extracting waterlogged corn based on satellite remote sensing. Background Art

[0002] Satellite remote sensing has the advantages of large range, low cost, objectivity and high efficiency, and is an effective means to quickly identify and extract the spatial distribution of waterlogged crops. At present, based on the principle of water remote sensing, it is relatively easy to accurately identify and extract water information because water has a spectral characteristic with significantly lower reflectivity than other land objects. Therefore, the known research on extracting waterlogged areas using remote sensing technology often focuses on the enhancement and extraction of water information.

[0003] However, in actual applications, the image features of flood-affected areas are different from those of water bodies, showing unique spatial complexity and heterogeneity. As for farmland flooding, in addition to the obvious water body characteristics, there are also characteristics of crops wilting due to long-term flooding or the coexistence of water bodies and crops in the flooded area. These situations have completely different characteristics in remote sensing images, which often lead to uncertainty in remote sensing identification of flooding. Previous studies have "treated flood-affected areas as water areas" for extraction, and have paid little attention to the problem of "spatial heterogeneity and complexity of farmland flood-affected areas", which is also the difficulty in quickly extracting flood-affected areas. Summary of the invention

[0004] In order to solve at least one technical problem existing in the prior art, the present application provides a method for extracting waterlogged corn based on satellite remote sensing.

[0005] The present application discloses a method for extracting waterlogged corn based on satellite remote sensing, comprising the following steps:

[0006] Step 1: Obtain field marker data in the target area, as well as the first optical remote sensing image data of the first period not affected by the waterlogging, and the second and third optical remote sensing image data of the two periods after the waterlogging;

[0007] Step 2: extracting the spatial distribution data of corn planting in the target area based on the first optical remote sensing image data;

[0008] Step 3: combining the field mark point data, respectively establishing identification marks of the waterlogged corn and the non-waterlogged corn in the second and third optical remote sensing image data;

[0009] Step 4: Combining the identification mark obtained in step 3, respectively obtain the spectral curves of the waterlogged corn and the non-waterlogged corn in the second and third optical remote sensing image data, and the change characteristics of the spectral curve of the waterlogged corn over time;

[0010] Step 5: spectral characteristic analysis is performed on the spectral curves of the waterlogged corn and the non-waterlogged corn to obtain the common spectral characteristics that distinguish the waterlogged corn from the non-waterlogged corn;

[0011] Step 6: construct a waterlogged corn extraction index relationship based on the common spectral features obtained in step 5;

[0012] Step 7: Calculate the flood-damaged corn extraction index of the second and third optical remote sensing image data using the flood-damaged corn extraction index relationship constructed in step 6, and then combine it with the corn planting spatial distribution data obtained in step 2 to obtain a binary distribution map of flood-damaged corn.

[0013] In an optional implementation, in step 1, all optical remote sensing image data are from Landsat 8 satellite image data, and all optical remote sensing image data contain the following 6 bands:

[0014] Blue, green, red, near infrared, shortwave infrared 1, shortwave infrared 2.

[0015] In an optional implementation, the field marker point data is obtained by real-time positioning using GPS, wherein the waterlogging and non-waterlogging attributes of corn in the target area are recorded.

[0016] In an optional implementation, in the step 2, a support vector machine method is used to extract the spatial distribution data of corn planting in the target area;

[0017] In addition, the step 2 further comprises:

[0018] The confusion matrix method is used to evaluate the accuracy of the obtained maize planting spatial distribution data. If the accuracy meets the requirements, it proceeds to the next step. If the accuracy does not meet the requirements, it returns to adjust the training samples in the support vector machine method until the accuracy meets the requirements.

[0019] In an optional embodiment, in step four, the reflectance of the pixels at corresponding positions on the second and third optical remote sensing image data are extracted according to the waterlogging and non-waterlogging attributes of the corn in the field marker point data, and the reflectance of multiple pixels is averaged to obtain the spectral curves of the waterlogged corn and the non-waterlogged corn. In addition, the average reflectance obtained on the third optical remote sensing image data is subtracted from the average reflectance obtained on the second optical remote sensing image data to obtain the temporal variation characteristics of the spectral curve of the waterlogged corn.

[0020] In an optional embodiment, in step 5, the common spectral characteristics of waterlogged corn and non-waterlogged corn are as follows:

[0021] 1) The reflectance of waterlogged corn in the near-infrared band was significantly lower than that of non-waterlogged corn, and the reflectance of red band was higher than that of non-waterlogged corn;

[0022] 2) As time goes by, the reflectivity of waterlogged corn increases from blue to short-wave infrared bands, while the reflectivity of non-waterlogged corn decreases.

[0023] In an optional implementation, in step 6, the following two waterlogged corn extraction index relationship equations are constructed based on the common spectral characteristics:

[0024] Ratio vegetation index: RVI = R i4 / R i3 (1);

[0025] Difference and index: DS = (R i1 -R j1 )+(R i2 -R j2 )+(R i3 -R j3 )+(R i4 -R j4 )+(R i5 -R j5 )+(R i6 -R j6 )(2);

[0026] Among them, R represents reflectivity, letter subscript i corresponds to the second optical remote sensing image data, letter subscript j corresponds to the third optical remote sensing image data, and numerical subscripts 1-6 correspond to the blue, green, red, near-red, short-wave infrared 1 and short-wave infrared 2 bands in the optical remote sensing image data, respectively.

[0027] In an optional implementation manner, the step seven specifically includes:

[0028] Calculating the ratio vegetation index of the second optical remote sensing image data, and considering the pixels therein that are lower than the first initial threshold as waterlogging areas, thereby obtaining a first waterlogging area;

[0029] Calculating the difference and index of the second and third optical remote sensing image data, and considering the pixels below the second initial threshold as the waterlogged area, thereby obtaining the second waterlogged area;

[0030] Merging the first flood-affected area with the second flood-affected area to obtain a final flood-affected area;

[0031] The final flood-affected area is masked using the spatial distribution data of corn planting obtained in step 2 to obtain a binary distribution map of flood-affected corn.

[0032] In an optional implementation, the second initial threshold is 0.

[0033] In an optional embodiment, the method for extracting waterlogged corn based on satellite remote sensing further comprises:

[0034] Step 8: vectorize the binary distribution map of waterlogged corn obtained in step 7, delete independent small patches of less than or equal to 2 pixels, and obtain the final binary distribution map of waterlogged corn.

[0035] This application has at least the following beneficial technical effects:

[0036] 1) This application is a method for extracting flooded corn based on the common characteristics of flooded corn remote sensing images and the changing characteristics of spectral curves. It fully considers the complexity and diversity of flooded corn image characteristics, making the extraction of flooded areas more comprehensive and the accuracy of flood remote sensing identification higher;

[0037] 2) In the method for extracting waterlogged corn based on satellite remote sensing of the present application, optical images after two periods of flooding are used to solve the problem that the image features of waterlogged corn change over time;

[0038] 3) In the method for extracting flood-affected corn based on satellite remote sensing of the present application, a technical method for extracting the flood-affected corn range by combining the ratio vegetation index (RVI) with the difference sum index (DS) is established based on the two common characteristics that the spectral characteristics of flood-affected corn are different from those of non-flood-affected corn. The constructed method makes up for the problem that a single index is difficult to fully extract the flood-affected area;

[0039] 4) The method for extracting waterlogged corn based on satellite remote sensing in the present application is simple and practical. The newly constructed DS index is to reveal that the reflectance change trend of waterlogged corn from blue to short-wave infrared bands is opposite to that of non-waterlogged corn. In specific applications, the remote sensing data source does not necessarily have to have all the bands involved in the present application. It only needs to have some bands and be constructed according to the idea of ​​DS index. Therefore, it has wider applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of the method for extracting waterlogged corn based on satellite remote sensing of the present application;

[0041] Figure 2 It is a schematic diagram of identification marks of waterlogged corn and non-waterlogged corn in the second optical remote sensing image data in a specific example of the present application (R: near infrared; G: short-wave infrared; B: red);

[0042] Figure 3 It is a schematic diagram of identification marks of waterlogged corn and non-waterlogged corn in the third optical remote sensing image data in a specific example of the present application (R: near infrared; G: short-wave infrared; B: red);

[0043] Figure 4is a reflectance curve of waterlogged corn and non-waterlogged corn in the second optical remote sensing image data in a specific example of the present application;

[0044] Figure 5 is a reflectance curve of waterlogged corn and non-waterlogged corn in the third optical remote sensing image data in a specific example of the present application;

[0045] Figure 6 is a graph showing the reflectivity variation over time of waterlogged corn and non-waterlogged corn obtained based on the second and third optical remote sensing image data in a specific example of the present application;

[0046] Figure 7 The spatial distribution map of flood-damaged corn and non-flood-damaged corn (ie, a binary distribution map) is obtained by applying the flood-damaged corn extraction method based on satellite remote sensing of the present application to a specific case. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the implementation of this application clearer, the technical scheme in the embodiment of this application will be described in more detail below in combination with the embodiment of this application and the accompanying drawings. The described embodiment is a part of the embodiment of this application, not all of the embodiments. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain this application, and should not be construed as limiting this application.

[0048] Based on the Landsat8 satellite data, this application studies the common spectral characteristics of waterlogged corn, and establishes a waterlogged corn identification method based on the complex and diverse image characteristics of waterlogged corn. It is intended to focus on solving the following three problems:

[0049] 1) Whether the common characteristics of remote sensing images of waterlogged corn in various situations and their spectral curves are different from those of normal corn, and how they change over time;

[0050] 2) How to construct a rapid extraction method for waterlogged corn based on the common characteristics of remote sensing images of waterlogged corn and the changing characteristics of spectral curves;

[0051] 3) The application of the constructed waterlogged corn identification method at the county level.

[0052] The following is an exemplary description of the present application in combination with the application test of the present application method in Suixi County, Anhui Province:

[0053] From July to September 2021, some areas of Anhui Province were affected by the passage of typhoons such as "Fireworks" and "Meihua". Multiple rounds of precipitation caused waterlogging in some farmland, seriously affecting the growth and development of corn. In some areas, corn was completely harvested. The method of this application was used to identify waterlogged corn in the area, including Figure 1 The following steps are described:

[0054] Step 1: Obtain field marker data within the target area, as well as the first optical remote sensing image data of the first phase that is not affected by the flood, and the two phases of optical remote sensing image data after being eroded by the flood, which are the second and third optical remote sensing image data.

[0055] Among them, the field landmark point data is obtained by real-time positioning with GPS, which records the waterlogging and non-waterlogging attributes of corn in the target area. These two attributes are used in subsequent steps to correspond to remote sensing images and establish identification marks for waterlogged corn.

[0056] The above-mentioned first optical remote sensing image data of the first period that is not affected by waterlogging is preferably data within the critical identification period of corn, which is used in subsequent steps to extract the spatial distribution map of corn planting (i.e., corn planting spatial distribution data); in this embodiment, considering that there is more cloud and rainy weather in the study area during the critical identification period of corn in 2021, in order to avoid the impact of clouds on the extraction of corn planting spatial distribution, the remote sensing image that is not affected by waterlogging here uses the Landsat8 data on August 28, 2020.

[0057] Furthermore, the two optical remote sensing image data after the flood erosion (i.e., the second and third optical remote sensing image data) are used in subsequent steps to distinguish between flooded and non-flooded corn. The time interval between them is preferably more than half a month, and the data dates are all within the corn growing period; in this embodiment, two Landsat8 data from July 30 and August 31, 2021 are used.

[0058] Furthermore, the above-mentioned Landsat8 images are all pre-processed surface reflectance products, which contain six bands commonly used in agricultural remote sensing monitoring, namely blue (0.45-0.51μm), green (0.53-0.59μm), red (0.64-0.67μm), near infrared (0.85-0.88μm), shortwave infrared 1 (1.57-1.65μm), and shortwave infrared 2 (2.11-2.29μm).

[0059] Step 2: Based on the first optical remote sensing image data, extract the spatial distribution data of corn planting in the target area.

[0060] Specifically, this step is based on the Landsat8 image on August 28, 2020, and uses the mature support vector machine method to extract the spatial distribution data of corn planting in the target area.

[0061] Furthermore, this step 2 may also include the following steps:

[0062] The classic confusion matrix method is used to evaluate the accuracy of the maize planting spatial distribution data obtained above. If the accuracy meets the requirements, proceed to the next step. If the accuracy does not meet the requirements, return to adjust the training samples in the support vector machine method until the accuracy meets the requirements.

[0063] Step 3: Combine the field landmark data to establish Figure 2 and Figure 3 Identification marks of waterlogged corn and non-waterlogged corn in the second and third optical remote sensing image data shown.

[0064] from Figure 2 , Figure 3 It can be seen that waterlogged corn has the following four features (i.e. identification marks, or image features) in the image:

[0065] 1) Similar water bodies (such as Figure 2 , Figure 3 1 corn site in moderate flooding);

[0066] 2) Water bodies and crops coexist (e.g. Figure 2 , Figure 3 2 places of corn in moderate flooding);

[0067] 3) The green hue in the image after the plants wilt due to waterlogging (such as Figure 2 , Figure 3 3 places for corn in moderate flooding);

[0068] 4) As the flooding duration increases, the flooded corn at locations 1 and 2 in the figure show bright color characteristics, and the image characteristics of the flooded corn at location 3 almost change.

[0069] In summary, it is shown that waterlogged corn has the image characteristics of spatial complexity and heterogeneity.

[0070] Step 4: Combine the identification marks obtained in step 3 to obtain the spectral curves of the flood-affected corn and the non-flood-affected corn in the second and third optical remote sensing image data, and the change characteristics of the spectral curves of the flood-affected corn over time.

[0071] Specifically, this step is to extract the reflectance of the pixels at the corresponding positions on the second and third optical remote sensing image data according to the waterlogging and non-waterlogging attributes of the corn in the field landmark data, and then average the reflectance of multiple pixels (that is, obtain the reflectance of waterlogged corn and non-waterlogged corn in different bands in the Landsat8 data on July 30, 2021, and the reflectance of waterlogged corn and non-waterlogged corn in different bands in the Landsat8 data on August 31, 2021), so as to obtain the following Figure 4 and Figure 5 The spectral curves of flooded corn and non-flooded corn are shown (the spectral curve here refers to the curve of the relationship between spectral wavelength and reflectivity).

[0072] In addition, the average reflectivity obtained on the second optical remote sensing image data is subtracted from the average reflectivity obtained on the third optical remote sensing image data to obtain the following: Figure 6 The graph showing the reflectance variation of flooded corn and non-flooded corn over time shows the variation characteristics of the spectral curve of flooded corn over time.

[0073] Step 5: Analyze the spectral characteristics of the spectral curves of the waterlogged corn and the non-waterlogged corn to obtain the common spectral characteristics that distinguish the waterlogged corn from the non-waterlogged corn.

[0074] Specifically, by comparing the spectral curves of flood-damaged corn and non-flood-damaged corn obtained in the above steps, two conclusions can be drawn:

[0075] 1) See Figure 4 and Figure 5 As shown, in the images taken on July 30 and August 31, 2021, the reflectivity of waterlogged corn (i.e., the affected corn in the picture) in the near-infrared band is significantly lower than that of non-waterlogged corn (i.e., the normal corn in the picture), and the reflectivity of the red band is higher than that of non-waterlogged corn;

[0076] 2) See Figure 6 As shown, over time, the reflectivity of waterlogged corn (i.e., the affected corn in the figure) increased from the blue to the short-wave infrared band, while the reflectivity of non-waterlogged corn (i.e., the normal corn in the figure) decreased.

[0077] In addition, since the spectral curve of waterlogged corn covers the pixels of the four image features (i.e., identification marks) mentioned in step three, these two conclusions are applicable to the four image features and are the common features (common spectral features) that distinguish waterlogged corn from non-waterlogged corn.

[0078] Step 6: Based on the common spectral features obtained in step 5, construct an index relationship (index combination) for extracting waterlogged corn.

[0079] Specifically, this step is to construct the following two waterlogging corn extraction index relationship equations based on the above two conclusions:

[0080] Ratio vegetation index (based on the conclusion in point 1 above): RVI = R i4 / R i3 (1);

[0081] Difference and index (based on the conclusion in point 2 above): DS = (R i1 -R j1 )+(R i2 -R j2 )+(R i3 -R j3 )+(Ri4 -R j4 )+(R i5 -R j5 )+(R i6 -R j6 )(2);

[0082] Among them, R represents reflectivity, letter subscript i corresponds to the second optical remote sensing image data, letter subscript j corresponds to the third optical remote sensing image data, and numerical subscripts 1-6 correspond to the blue, green, red, near-red, short-wave infrared 1 and short-wave infrared 2 bands in the optical remote sensing image data, respectively.

[0083] Step 7: Calculate the flood-damaged corn extraction index of the second and third optical remote sensing image data using the flood-damaged corn extraction index relationship constructed in step 6, and then combine it with the corn planting spatial distribution data obtained in step 2 to obtain a binary distribution map of flood-damaged corn.

[0084] Specifically, this step seven includes:

[0085] Step 7.1: Use the above relationship / formula (1) to calculate the ratio vegetation index (RVI) of the second optical remote sensing image data on July 30, 2021. In this example, the RVI is 8.5, and set it as the first initial threshold. Figure 4 , Figure 5 The spectral curve of waterlogged corn has a lower RVI than normal corn. Therefore, the pixels below the first initial threshold are regarded as waterlogged areas, thus obtaining the first waterlogged area.

[0086] Step 7.2: Use the above relationship / formula (2) to calculate the difference and index (DS) of the second and third optical remote sensing image data on July 30 and August 31, according to Figure 6 It can be seen that the DS value of waterlogged corn should be negative, and the DS value of non-waterlogged corn should be positive. Therefore, 0 is used as the second initial threshold of DS, and the pixels below the second initial threshold are regarded as waterlogged areas, thereby obtaining the second waterlogged area.

[0087] Step 7.3: Merge the first flood-affected area with the second flood-affected area to obtain the final flood-affected area.

[0088] Step 7.4: Mask the final flood-affected area using the spatial distribution data of corn planting obtained in step 2 to obtain a binary distribution map of flood-affected corn.

[0089] Furthermore, the method for extracting waterlogged corn based on satellite remote sensing of the present application may also include the following steps:

[0090] Step 8. Vectorize the binary distribution map of waterlogged corn obtained in step 7 (vectorization here refers to the operation of converting raster to vector, which is implemented in remote sensing software), delete independent small patches of less than or equal to 2 pixels, and obtain the final binary distribution map of waterlogged corn (i.e., spatial distribution data).

[0091] Based on this method, the flood-damaged corn area in Suixi County is 613,300 mu, and the flood-damaged corn area (flood-damaged corn area divided by corn planting area) is 29.11%. Figure 7 The spatial distribution map of normal corn in Figure 7 shown.

[0092] Finally, in order to verify the feasibility of the method proposed in this application, the extraction accuracy of waterlogged corn was calculated based on the field survey sample data using the following formula (3):

[0093]

[0094] Among them, P represents the extraction accuracy of waterlogged corn based on satellite data, and N1 and N2 represent the number of correctly identified sample points and the total number of sample points at the corresponding position, respectively.

[0095] Among the 129 verification points obtained based on the field survey, 121 verification points were correctly extracted as the affected corn layer. According to formula (3), the extraction accuracy of flood-affected corn in Suixi County is 94%. This shows that the technical method of the present application is feasible, and remote sensing technology can be used as an effective means to quickly extract the spatial distribution range of flood-affected corn.

[0096] It is specially noted that due to the high amount of cloud and rain in Suixi County during the critical corn identification period in 2021, the above-mentioned Landsat8 data on August 28, 2020 is used to extract the spatial distribution map of corn planting. Of course, if the technical method of this application is used in other areas, it is better to use optical images in the year of the disaster and during the critical corn identification period, if available, to avoid the problem of interannual variation in corn planting.

[0097] In summary, from the application effect point of view, the waterlogged corn extraction method based on satellite remote sensing in this application has achieved good results in the waterlogged corn monitoring in Suixi County, Anhui Province, my country, and has the advantages of simple calculation and fast identification, and has the advantage of further promotion and application.

[0098] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for extracting waterlogged corn based on satellite remote sensing, characterized in that: The steps include: Step 1: Obtain field marker data in the target area, as well as the first optical remote sensing image data of the first period not affected by the waterlogging, and the second and third optical remote sensing image data of the two periods after the waterlogging; Step 2: extracting the spatial distribution data of corn planting in the target area based on the first optical remote sensing image data; Step 3: combining the field mark point data, respectively establishing identification marks of the waterlogged corn and the non-waterlogged corn in the second and third optical remote sensing image data; Step 4: Combining the identification mark obtained in step 3, respectively obtain the spectral curves of the waterlogged corn and the non-waterlogged corn in the second and third optical remote sensing image data, and the change characteristics of the spectral curve of the waterlogged corn over time; Step 5, performing spectral characteristic analysis on the spectral curves of the waterlogged corn and the non-waterlogged corn to obtain the common spectral characteristics that distinguish the waterlogged corn from the non-waterlogged corn; Step 6: Based on the common spectral characteristics obtained in step 5, construct a waterlogged corn extraction index relationship; Step 7: Calculate the waterlogged corn extraction index of the second and third optical remote sensing image data by using the waterlogged corn extraction index relational expression constructed in step 6, and then combine it with the corn planting spatial distribution data obtained in step 2 to obtain a binary distribution map of waterlogged corn; The step seven specifically includes: Calculating the ratio vegetation index of the second optical remote sensing image data, and considering the pixels therein that are lower than the first initial threshold as waterlogging areas, thereby obtaining a first waterlogging area; Calculating the difference and index of the second and third optical remote sensing image data, and considering the pixels below the second initial threshold as the waterlogged area, thereby obtaining the second waterlogged area; Merging the first flood-affected area with the second flood-affected area to obtain a final flood-affected area; The final flood-affected area is masked using the spatial distribution data of corn planting obtained in step 2 to obtain a binary distribution map of flood-affected corn.

2. The method for extracting waterlogged corn based on satellite remote sensing according to claim 1, characterized in that: In step 1, all optical remote sensing image data are from Landsat8 satellite image data, and all optical remote sensing image data contain the following 6 bands: Blue, green, red, near infrared, shortwave infrared 1, shortwave infrared 2.

3. The method for extracting waterlogged corn based on satellite remote sensing according to claim 1, characterized in that: The field marker point data are obtained by real-time positioning using GPS, and record the waterlogging and non-waterlogging attributes of corn in the target area.

4. The method for extracting waterlogged corn based on satellite remote sensing according to claim 1, characterized in that: In the step 2, a support vector machine method is used to extract the spatial distribution data of corn planting in the target area; In addition, the step 2 further comprises: The confusion matrix method is used to evaluate the accuracy of the obtained maize planting spatial distribution data. If the accuracy meets the requirements, it proceeds to the next step. If the accuracy does not meet the requirements, it returns to adjust the training samples in the support vector machine method until the accuracy meets the requirements.

5. The method for extracting waterlogged corn based on satellite remote sensing according to claim 1, characterized in that: In the step four, the reflectance of the pixels at the corresponding positions on the second and third optical remote sensing image data are extracted according to the waterlogging and non-waterlogging attributes of the corn in the field marker point data, and the reflectance of multiple pixels is averaged to obtain the spectral curves of the waterlogged corn and the non-waterlogged corn. In addition, the average reflectance obtained on the second optical remote sensing image data is subtracted from the average reflectance obtained on the third optical remote sensing image data to obtain the change characteristics of the spectral curve of the waterlogged corn over time.

6. The method for extracting waterlogged corn based on satellite remote sensing according to claim 2, characterized in that: In step 5, the common spectral characteristics of waterlogged corn and non-waterlogged corn are as follows: 1) The reflectance of waterlogged corn in the near-infrared band was significantly lower than that of non-waterlogged corn, and the reflectance of red band was higher than that of non-waterlogged corn; 2) As time goes by, the reflectivity of waterlogged corn increases from blue to short-wave infrared bands, while the reflectivity of non-waterlogged corn decreases.

7. The method for extracting waterlogged corn based on satellite remote sensing according to claim 6, characterized in that: In step six, the following two waterlogged corn extraction index relationship equations are constructed based on the common spectral characteristics: Ratio vegetation index: RVI = R i4 / R i3 (1); Difference and index: DS = (R i1 -R j1 )+(R i2 -R j2 )+(R i3 -R j3 )+(R i4 -R j4 )+(R i5 -R j5 )+(R i6 -R j6 )(2); Among them, R represents reflectivity, letter subscript i corresponds to the second optical remote sensing image data, letter subscript j corresponds to the third optical remote sensing image data, and numerical subscripts 1-6 correspond to the blue, green, red, near-red, short-wave infrared 1 and short-wave infrared 2 bands in the optical remote sensing image data, respectively.

8. The method for extracting waterlogged corn based on satellite remote sensing according to claim 7, characterized in that: The second initial threshold is 0.

9. The method for extracting waterlogged corn based on satellite remote sensing according to claim 7, characterized in that: Also includes: Step 8: vectorize the binary distribution map of waterlogged corn obtained in step 7, delete independent small patches of less than or equal to 2 pixels, and obtain the final binary distribution map of waterlogged corn.

Images