Agriculture and forestry monitoring and early warning method and system based on digital model and remote sensing

Abstract

The present application relates to the technical field of agricultural digital monitoring and early warning, in particular to a forestry monitoring and early warning method and system based on digital models and remote sensing, mainly comprising the following steps: segmenting and extracting a plurality of field regions from remote sensing image data, marking a first planting region greater than a planting trend rate threshold and a second planting region not greater than the planting trend rate threshold, establishing an evaluation model for the second planting region, sorting the evaluation values, retaining the maturity of the N second planting regions arranged from large to small in the sorting result, and replacing the maturity of the remaining second planting regions with the maturity of the first planting region to form a third planting region. Through the above method, the entire planting region is divided into a plurality of defined different planting regions, the situation of each planting region is comprehensively analyzed and evaluated, a relatively reasonable early warning judgment value is obtained, and a reasonable early warning level is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural digital monitoring and early warning, in particular to a forestry monitoring and early warning method and system based on digital models and remote sensing. BACKGROUND

[0002] In the field of agriculture and forestry, the use of digital models for monitoring reduces the impact of manual activities on the environment and attempts to achieve early warning detection under a zero-carbon (carbon emission reduction) state. The main point is to use cloud computing, big data, remote sensing, GIS, and other modern information technologies under the support and constraints of unified operation and maintenance management, technical standards, and security systems, in accordance with unified mathematical benchmarks, to achieve the standardized integration, database construction, and centralized display of decentralized, multi-source, and heterogeneous agricultural geographic spatial data. Geographic information services are the core, and the open sharing of agricultural geographic spatial data is provided to various agricultural application systems to provide unified geographic spatial data and functional service support. Through the organic superposition of industry integration data, map-based land management and map-based agriculture are achieved to provide spatial data support and decision support for the optimal allocation and macro-control of agricultural resources.

[0003] However, the current early warning monitoring system generally analyzes and evaluates the entire planting area uniformly and then warns of the possible loss degree of the current planting area. According to the early warning signals of different loss degrees, local personnel will take different measures to achieve better results under appropriate measures. However, when there are different types of crops in the entire planting area, uniform evaluation may result in unreasonable early warning results, causing the corresponding measures to be incompatible with the early warning results, resulting in high or far from enough management costs. SUMMARY

[0004] The purpose of the present application is to provide a forestry monitoring and early warning method and system based on digital models and remote sensing to solve the above problems in the prior art.

[0005] The present application is achieved by the following technical solutions:

[0006] In a first aspect, the present application provides a forestry monitoring and early warning method based on digital models and remote sensing, comprising:

[0007] Obtaining remote sensing image data, performing land division and extraction on the remote sensing image data to obtain a plurality of field regions, identifying crops in the field regions to obtain the current growth area of a plurality of crops;

[0008] Obtaining historical average crop growth area, obtaining current crop planting trend rate through historical average crop growth area and current growth area, setting planting trend rate threshold, marking first planting area greater than planting trend rate threshold and second planting area not greater than planting trend rate threshold;

[0009] Obtaining current image data of first planting area and second planting area and a plurality of stage different maturity contrast images, obtaining maturity of first planting area and maturity of second planting area based on current image data and marked different maturity contrast images;

[0010] Obtaining maturity of first planting area adjacent to second planting area, establishing evaluation model of second planting area, outputting evaluation value of current second planting area based on maturity of first planting area adjacent to second planting area and evaluation model;

[0011] Sorting evaluation value, setting selection threshold N, retaining maturity of N second planting areas arranged from large to small in sorting result, replacing maturity of remaining second planting areas with maturity of first planting area, forming third planting area;

[0012] Obtaining current meteorological data, establishing early warning model, outputting early warning level through early warning model based on meteorological data, maturity of first planting area, maturity of second planting area and maturity of third planting area.

[0013] Preferably, the step of segmenting and extracting the remote sensing image data to obtain a plurality of field block regions comprises:

[0014] Obtaining historical images with field blocks and field block edges for labeling, establishing image recognition model, training the image recognition model through historical images, obtaining trained image recognition model;

[0015] Extracting field block edge features based on the image recognition model, and segmenting the current field block into a plurality of field block regions based on the field block edge features.

[0016] Preferably, the step of segmenting the current field block into a plurality of field blocks based on the field block edge feature information comprises:

[0017] Judging whether the field block region is a closed figure, wherein the closed figure is that the entire field block region is surrounded by the field block edge feature;

[0018] If the field block region is a closed figure, the current field block region area is obtained, the historical average field block region area is obtained, and the error area is set. When the absolute value of the difference between the current field block region area and the historical average field block region area is greater than the error area, a first signal of current target field block region recognition error is output, otherwise, a signal of current target field block region recognition normal is output.

[0019] If the field block region is a non-closed figure, a second signal of current target field block region recognition error is output.

[0020] The target field block region is re-identified according to the first signal and the second signal respectively.

[0021] Preferably, the re-identification of the target field block region according to the first signal and the second signal respectively comprises:

[0022] When the first signal of current target field block region recognition error is received, the highest and lowest elevation points of the target field block region are extracted, the elevation difference is obtained, and the elevation difference threshold is set. If the elevation difference is greater than the elevation difference threshold, the current target field block region is deleted, otherwise, the target field block region is output as a normal field block region.

[0023] When the second signal of current target field block region recognition error is received, a plurality of endpoints of the non-closed figure are obtained, the closest two endpoints are connected to obtain a target line segment, the length of the target line segment and the length of the edge feature of the target field block region are obtained to obtain a proportion value, and a proportion threshold is set. If the proportion value is less than the proportion threshold, the target field block region is output as a normal field block region, otherwise, the current target field block region is deleted.

[0024] Preferably, the planting trend rate of the current crop is obtained by the historical crop growth area and the current growth area, and a planting trend rate threshold is set.

[0025]

[0026] In the formula, is the planting trend rate, is the historical average growth area of the current crop, is the current growth area of the current crop.

[0027] Preferably, the maturity of the first planting area and the maturity of the second planting area are obtained based on the current image data and the contrast image marked with different maturities.

[0028] The crop image features in the extracted current image data and contrast image are extracted, and the target RGB value in the crop image feature and the contrast RGB value of the crop in the plurality of contrast images are extracted respectively.

[0029] The matching degree model is established, the matching degree is calculated based on the target RGB value and the contrast RGB value through the matching degree model, the contrast image with the maximum matching degree is selected as the target image, and the maturity of the target image is stored as the maturity of the current image data.

[0030] Preferably, the matching degree model comprises:

[0031]

[0032] In the formula, is the matching degree, is the red luminance value of the current image data, is the red luminance value of the i-th contrast image, is the red luminance value of the i-th contrast image, is the red luminance value of the current image data, is the green luminance value of the i-th contrast image, is the green luminance value of the i-th contrast image, is the blue luminance value of the current image data, is the blue luminance value of the i-th contrast image.

[0033] Preferably, the evaluation model of the second planting area comprises:

[0034]

[0035] In the formula, is the evaluation value of the i-th second planting area, is the numerical value of the type of the first crop adjacent to the current second planting area, is the number of the first planting areas adjacent to the current second planting area, is the area of the o-th first planting area adjacent to the current second planting area, is the maturity of the crop in the o-th first planting area adjacent to the current second planting area, .

[0036] Preferably, the maturity of the remaining second planting areas is replaced by the maturity of the first planting area.

[0037] The maturity of the first planting area adjacent to the remaining second planting area is selected as the maturity of the third planting area;

[0038] The early warning model comprises:

[0039]

[0040] In the formula, is the early warning judgment value,​​ an average maturity of crops in a first planting area, an average maturity of crops in a second planting area, an average maturity of crops in a third planting area, an average wind speed throughout the year, a current wind speed, an average rainfall suitable for crop growth, a current rainfall;

[0041] a plurality of early warning thresholds are set, and a current early warning level is output according to the early warning judgment value and the early warning threshold.

[0042] In a second aspect, the present application also provides a method for monitoring and early warning of agriculture and forestry based on a digital model and remote sensing, comprising:

[0043] a field identification module configured to obtain remote sensing image data, perform field segmentation on the remote sensing image data to extract a plurality of field areas, identify crops in the field areas to obtain current growth areas of a plurality of crops, obtain historical average growth areas of the crops, obtain a planting trend rate of the current crops from the historical average growth areas of the crops and the current growth areas, and set a planting trend rate threshold to mark a first planting area greater than the planting trend rate threshold and a second planting area not greater than the planting trend rate threshold;

[0044] an early warning judgment module configured to obtain current image data of the first planting area and the second planting area and a plurality of control images of different maturities, obtain a maturity of the first planting area and a maturity of the second planting area based on the current image data and the control images marked with different maturities, obtain a maturity of the first planting area adjacent to the second planting area, establish an evaluation model of the second planting area, output an evaluation value of the current second planting area based on the maturity of the first planting area adjacent to the second planting area and the evaluation model, sort the evaluation values, set a selection threshold N, retain the maturities of the N second planting areas arranged from large to small in the sorting result, replace the maturities of the remaining second planting areas with the maturity of the first planting area to form a third planting area, obtain current meteorological data, establish an early warning model, and output an early warning level based on the meteorological data, the maturity of the first planting area, the maturity of the second planting area, and the maturity of the third planting area through the early warning model;

[0045] a master control device connected with the field identification module and the early warning judgment module, configured to execute the method for monitoring and early warning of agriculture and forestry based on the digital model and the remote sensing.

[0046] The technical solution of the present application has at least the following advantages and beneficial effects:

[0047] The method provided by the application mainly comprises the following steps: performing field block segmentation on remote sensing image data to obtain a plurality of field block regions, marking a first planting region greater than a planting trend rate threshold and a second planting region not greater than the planting trend rate threshold, establishing an evaluation model of the second planting region, sorting the evaluation values, retaining the maturity of N second planting regions arranged from large to small in the sorting result, and replacing the maturity of the remaining second planting regions with the maturity of the first planting region to form a third planting region. Through the above method, the entire planting region is divided into a plurality of planting regions defined as different planting regions. Through comprehensive analysis and evaluation of the condition of each planting region, a relatively reasonable early warning judgment value is obtained, and a reasonable early warning level is obtained, thereby providing an accurate judgment signal for the staff and making reasonable subsequent measures. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0049] Fig. 1 The control flow diagram of the present application;

[0050] Fig. 2 The system structure diagram of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical scheme in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0052] The terms "first", "second" and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily describe a specific order or a logical sequence. The naming or numbering of the steps appearing in the present application does not mean that the steps in the method flow must be executed in the order indicated by the naming or numbering. The execution order of the named or numbered flow steps can be changed according to the technical purpose to be achieved, as long as the same or similar technical effect can be achieved.

[0053] Please refer to Figs. 1-2The application provides a kind of agriculture and forestry monitoring and early warning method based on digital model and remote sensing, comprising:

[0054] S101: obtaining remote sensing image data, obtaining a plurality of field regions by field segmentation extraction on the remote sensing image data, and identifying crops in the field regions to obtain the current growth area of a plurality of crops;

[0055] Among them, satellite remote sensing is the main means to obtain large-scale ground information, common data sources include optical satellites (such as Landsat, Sentinel-2): provide multispectral and high-resolution images, suitable for land cover, vegetation monitoring, etc., radar satellites (such as Sentinel-1): can penetrate clouds, suitable for all-weather monitoring, such as geological and water resource survey; hyperspectral satellite: provides continuous spectral information, suitable for material composition analysis, in the application, the appropriate way is selected to obtain, the entire region is divided into a plurality of field regions to facilitate subsequent analysis of different regions, and the modular structure facilitates data labeling and differentiation from other regions.

[0056] S102: obtaining the historical average growth area of crops, obtaining the planting trend rate of the current crops by the historical average growth area of crops and the current growth area, and setting a planting trend rate threshold, marking the first planting area greater than the planting trend rate threshold and the second planting area not greater than the planting trend rate threshold;

[0057] In the application, the planting trend rate is selected to reflect the importance of the current crops, which roughly reflects the planting degree of the current crops each year, i.e., gradually increasing or gradually decreasing, so that this data becomes one of the subsequent judgment factors for early warning, the focus is on distinguishing crops with different planting trend rates and making different considerations in subsequent judgments, making the judgment more reasonable, and secondly, the crops in this embodiment can be agricultural crops or forestry crops.

[0058] S103: obtaining the current image data of the first planting area and the second planting area and a plurality of stage different maturity contrast images, obtaining the maturity of the first planting area and the maturity of the second planting area based on the current image data and the contrast images marked with different maturity;

[0059] In this embodiment, it is mainly used to judge the maturity of the current crop, and the maturity is selected as one of the other judgment factors for early warning in this embodiment, which means that the crop will be harvested soon, and the early warning level should be further improved when weather disasters occur, in this embodiment, different maturity is stored in advance for different contrast images, the maturity value ranges from 0 to 1, and the maturity of each contrast image can be defined according to the experience of those skilled in the art.

[0060] S104: Obtain the maturity of the first planting area adjacent to the second planting area, establish an evaluation model of the second planting area, and output an evaluation value of the current second planting area based on the maturity of the first planting area adjacent to the second planting area and the evaluation model;

[0061] S105: Sort the evaluation values, set a selection threshold N, and retain the maturity of the N second planting areas arranged from large to small in the sorting result, and replace the maturity of the remaining second planting areas with the maturity of the first planting area to form a third planting area;

[0062] Since the crops in the second planting area do not reach the planting trend rate threshold set in the embodiment, it means that the area of the first crop is increasing in the scheme but does not reach the set growth requirement, or has been decreasing. In this case, the crops in the second planting area with a low evaluation value are more unstable and may be replaced or sharply reduced at any time, so the maturity of the crops in the second planting area does not have a large reference proportion. If a certain data of the second crop has a large influence on the entire result, it will cause a large judgment error in the subsequent process, so the replacement is considered to improve the accuracy of the subsequent judgment.

[0063] S106: Obtain current weather data, establish a warning model, and output a warning level through the warning model based on the weather data, the maturity of the first planting area, the maturity of the second planting area, and the maturity of the third planting area.

[0064] The method provided by the application mainly includes obtaining remote sensing image data, performing field segmentation on the remote sensing image data to obtain a plurality of field regions, marking the first planting area greater than the planting trend rate threshold and the second planting area not greater than the planting trend rate threshold, replacing the maturity of the remaining second planting area with the maturity of the first planting area to form a third planting area. Through the above method, the entire planting area is divided into a plurality of regions defined as different crops, and the planting area is comprehensively analyzed and evaluated based on the condition of each planting area to obtain a relatively reasonable warning judgment value and a reasonable warning level.

[0065] In an example embodiment of the application, the field segmentation of the remote sensing image data to obtain a plurality of field regions comprises:

[0066] Obtain a historical image with field and field edge for labeling, establish an image recognition model, train the image recognition model through the historical image, and obtain a trained image recognition model;

[0067] The field plot edge features are extracted from remote sensing image data based on an image recognition model, and the current field plot is segmented into a plurality of field plot regions based on the field plot edge features.

[0068] Specifically, the field plot boundary vectors are mostly obtained by digitizing satellite images, cadastral data or topological maps. The high-resolution images are digitized by professional visual interpretation. Prior knowledge is introduced based on the consideration of pixel proximity, which can more accurately extract the field plot boundary. However, when obtaining large-scale field plot boundary data, the manual digitization method will consume a lot of time and is low in efficiency.

[0069] Deep learning can realize automatic field plot extraction. Deep image segmentation models such as Deeplabv3, UNet and Mask RCNN can perform semantic segmentation on remote sensing images, and divide them into individual field plot objects with semantic labels. Since semantic segmentation focuses on the compliance (accuracy) of the predicted image and the label image pixels, it ignores the edge information between the field plot objects, so the phenomenon of inaccurate or missing field plot object edges may occur. Field plot segmentation needs to divide adjacent field plots according to the edges into individual field plots, and the loss of edge information will greatly affect the labeling accuracy of the field plot internal pixels. Therefore, this research aims to extract the natural boundaries of the field plots and further extract the internal boundaries of the field plots, so that the field plot segmentation results are more accurate and more practical.

[0070] Satellite remote sensing technology is used to realize the identification of cultivated land field plots in administrative regions, and to provide statistical information of cultivated land field plots in each county and associated display of field plot attribute information including field plot crop type, growth status, soil type, meteorological data, etc.

[0071] In an example embodiment of the present application, the segmentation of the current field plot into a plurality of field plots based on the field plot edge feature information includes:

[0072] It is judged whether the field plot region is a closed figure, and the closed figure is a field plot edge feature surrounding the entire field plot region. If the field plot region is a closed figure, the area of the current field plot region is obtained, the historical average field plot region area is obtained, and the error area is set. If the absolute value of the difference between the current field plot region area and the historical average field plot region area is greater than the error area, a first signal indicating that the current target field plot region recognition is incorrect is output, otherwise, a signal indicating that the current field plot region recognition is normal is output. If the field plot region is a non-closed figure, a second signal indicating that the current target field plot region recognition is incorrect is output. The target field plot region is re-identified and judged according to the first signal and the second signal, respectively.

[0073] Specifically, the re-identification and judgment of the target field plot region according to the first signal and the second signal respectively includes:

[0074] When receiving the first signal of the current target field block area identification error, the highest and lowest elevation points of the target field block area are extracted, the elevation difference is obtained, the elevation difference threshold is set, if the elevation difference is greater than the elevation difference threshold, the current target field block area is deleted, otherwise, the target field block area is output as a normal field block area;

[0075] When receiving the second signal of the current target field block area identification error, a plurality of endpoints of the non-closed figure are obtained, the two closest endpoints are connected to obtain a target line segment, the length of the target line segment and the length of the edge feature of the target field block area are obtained to obtain a proportion value, the proportion threshold is set, if the proportion value is less than the proportion threshold, the target field block area is output as a normal field block area, otherwise, the current target field block area is deleted.

[0076] Through the above method, it is further judged whether the current identified field block area exists misidentification, and then the subsequent judgment error is reduced.

[0077] In an example embodiment of the present application, the planting trend rate of the current crop is obtained by the historical crop growth area and the current growth area, and the planting trend rate threshold includes:

[0078]

[0079] In the formula, The planting trend rate is The historical average growth area of the current crop is The current growth area of the current crop is The planting trend rate threshold is The numerical value of the crop type is The average trend rate of all crops is

[0080] In the present embodiment, the planting trend rate threshold can be set according to experience, which can be set to 0 in the present embodiment.

[0081] In an example embodiment of the present application, the maturity of the first planting area and the maturity of the second planting area are obtained based on the current image data and the reference image marked with different maturities, which includes:

[0082] The crop image features in the extracted current image data and reference image are extracted, the target RGB value in the crop image feature and the reference RGB value of the crop in the plurality of reference images are extracted respectively;

[0083] A matching degree model is established, the matching degree is calculated based on the target RGB value and the reference RGB value through the matching degree model, the reference image with the maximum matching degree is selected as the target image, and the maturity of the target image is stored as the maturity of the current image data.

[0084] Specifically, the matching degree model is established by:

[0085]

[0086] In the formula, is the matching degree, is the red luminance value of the current image data, is the red luminance value of the i-th reference image, is the red luminance value of the current image data, is the green luminance value of the i-th reference image, is the green luminance value of the current image data, is the blue luminance value of the i-th reference image, is the blue luminance value of the current image data, is the blue luminance value of the i-th reference image. In the embodiment, the crops are mainly wheat and rice, and different states have a large color difference, so the maturity of the crops is determined by the chroma, which is a faster solution. Of course, image recognition, feature extraction, and similarity comparison can also be used, but this method has a high cost and requires high computing power.

[0087] In an example embodiment of the present application, the evaluation model of the second planting area is established by:

[0088]

[0089] In the formula,

[0090] is the evaluation value of the i-th second planting area, is the value of the type of the first crop adjacent to the current second planting area, is the number of the first planting areas adjacent to the current second planting area, is the area of the o-th first planting area adjacent to the current second planting area, is the maturity of the crop in the o-th first planting area adjacent to the current second planting area, . In the embodiment, the higher the evaluation value of the second planting area, the more reference value it has. Subsequently, all evaluation values are sorted, and the front second planting areas are retained, and the rest are used as third planting areas. The data of the third planting area is replaced by the data of the adjacent first planting area.

[0091] In an example embodiment of the present application, the maturity of the rest of the second planting area is replaced by the maturity of the first planting area.

[0092] In an example embodiment of the present application, the maturity of the rest of the second planting area is replaced by the maturity of the first planting area. ​

[0093] arbitrarily selecting the maturity of the first planting area adjacent to the remaining second planting area as the maturity of the third planting area;

[0094] establishing a warning model includes:

[0095]

[0096] wherein, is a warning judgment value, is the average maturity of the crops in the first planting area, is the average maturity of the crops in the second planting area, is the average maturity of the crops in the third planting area, is the average wind speed throughout the year, is the current wind speed, is the average rainfall suitable for crop growth, is the current rainfall;

[0097] a number of warning thresholds are set, and the current warning level is output according to the warning judgment value and the warning threshold.

[0098] Two warning thresholds can be set, divided into three warning levels, and the warning level is output according to the range in which the warning judgment value falls.

[0099] A digital model and remote sensing-based agroforestry monitoring and warning method includes:

[0100] The field plot identification module is configured to obtain remote sensing image data, perform field plot segmentation on the remote sensing image data to extract a plurality of field plot areas, identify crops in the field plot areas to obtain current growth areas of a plurality of crops, obtain historical average growth areas of crops, obtain a planting trend rate of the current crops from the historical average growth areas of crops and the current growth areas, and set a planting trend rate threshold to mark a first planting area greater than the planting trend rate threshold and a second planting area not greater than the planting trend rate threshold.

[0101] The early warning judgment module is configured to acquire current image data of the first planting area and the second planting area and a plurality of stage different maturity contrast images, obtain the maturity of the first planting area and the maturity of the second planting area based on the current image data and the contrast images marked with different maturity, acquire the maturity of the first planting area adjacent to the second planting area, establish an evaluation model of the second planting area, output an evaluation value of the current second planting area based on the maturity of the first planting area adjacent to the second planting area and the evaluation model, sort the evaluation value, set a selection threshold N, retain the maturity of N second planting areas arranged from large to small in the sorting result, replace the maturity of the remaining second planting areas with the maturity of the first planting area to form a third planting area, acquire current meteorological data, establish an early warning model, and output an early warning level through the early warning model based on the meteorological data, the maturity of the first planting area, the maturity of the second planting area and the maturity of the third planting area.

[0102] A master control device is connected with the field identification module and the early warning judgment module, and is used for executing the above-mentioned digital model and remote sensing based agricultural and forestry monitoring and early warning method.

[0103] The above only is the preferred embodiment of the present application, and does not limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for monitoring and early warning of agriculture and forestry based on digital model and remote sensing, characterized in that, The method comprises the following steps: acquiring remote sensing image data, performing field segmentation on the remote sensing image data to obtain a plurality of field regions, identifying crops in the field regions, and obtaining current growth areas of a plurality of crops; acquiring historical average growth areas of the crops, obtaining a planting trend rate of the current crops based on the historical average growth areas and the current growth areas, setting a planting trend rate threshold, marking a first planting region greater than the planting trend rate threshold and a second planting region not greater than the planting trend rate threshold; acquiring current image data of the first planting region and the second planting region and a plurality of contrast images of different maturity stages, obtaining maturity of the first planting region and maturity of the second planting region based on the current image data and the contrast images marked with different maturity stages; acquiring the maturity of the first planting region adjacent to the second planting region, establishing an evaluation model of the second planting region, and outputting an evaluation value of the current second planting region based on the maturity of the first planting region adjacent to the second planting region and the evaluation model; sorting the evaluation values, setting a selection threshold N, retaining the maturity of the N second planting regions arranged from large to small in the sorting result, and replacing the maturity of the remaining second planting regions with the maturity of the first planting region to form a third planting region; acquiring current weather data, establishing a warning model, and outputting a warning level based on the weather data, the maturity of the first planting region, the maturity of the second planting region, and the maturity of the third planting region through the warning model; The method for obtaining a plurality of field regions by performing field segmentation on remote sensing image data comprises the following steps: acquiring historical images with field boundaries for labeling, establishing an image recognition model, training the image recognition model through the historical images, and obtaining a trained image recognition model; extracting field boundary features based on the image recognition model, and segmenting a current field into a plurality of field regions based on the field boundary features; The method for segmenting a current field into a plurality of fields based on field boundary feature information comprises the following steps: determining whether the field region is a closed figure, wherein the closed figure is a field region entirely surrounded by field boundary features; if the field region is a closed figure, acquiring a current field region area, acquiring a historical average field region area, setting an error area, outputting a first signal indicating that the current target field region is incorrectly identified when the absolute value of the difference between the current field region area and the historical average field region area is greater than the error area, otherwise, outputting a signal indicating that the current field region is correctly identified; if the field region is a non-closed figure, outputting a second signal indicating that the current target field region is incorrectly identified; re-identifying and determining the target field region according to the first signal and the second signal respectively; The method for re-identifying and determining the target field region according to the first signal and the second signal respectively comprises the following steps: When receiving the first signal of the current target field block area identification error, the highest and lowest points of the target field block area are extracted to obtain the height difference, and the height difference threshold is set. If the height difference is greater than the height difference threshold, the current target field block area is deleted, otherwise, the target field block area is output as a normal field block area; When receiving the second signal of the current target field block area identification error, a plurality of endpoints of the non-closed figure are obtained, the two closest endpoints are connected to obtain a target line segment, the length of the target line segment and the length of the edge feature of the target field block area are obtained to obtain a proportion value, and the proportion threshold is set. If the proportion value is less than the proportion threshold, the target field block area is output as a normal field block area, otherwise, the current target field block area is deleted; The planting trend rate of the current crop is obtained by the historical crop growth area and the current growth area, and the planting trend rate threshold is set, including: In the formula, is a planting trend rate, is a historical average growth area of the current crop, is a current growth area of the current crop; The maturity of the first planting area and the maturity of the second planting area are obtained based on the current image data and the control image marked with different maturities, including: The crop image features in the extracted current image data and control image are extracted, and the target RGB value in the crop image feature and the control RGB value of the crop in the plurality of control images are extracted respectively; A matching degree model is established, and the matching degree is calculated based on the target RGB value and the control RGB value through the matching degree model. The control image with the maximum matching degree is selected as the target image, and the maturity of the target image is stored as the maturity of the current image data; The maturity of the first planting area is selected to replace the maturity of the remaining second planting area, including: The maturity of the first planting area adjacent to the remaining second planting area is selected as the maturity of the third planting area; The establishment of the early warning model includes: wherein, is a warning decision value, is an average maturity of the crop in the first planting area, is an average maturity of the crop in the second planting area, is an average maturity of the crop in the third planting area, is an average wind speed for the year, is a current wind speed, is an average rainfall suitable for crop growth, is a current rainfall. A plurality of early warning thresholds are set, and the current early warning level is output according to the early warning judgment value and the early warning threshold.

Images