Rice plot identification method, device, equipment and medium

Abstract

The invention discloses a rice plot identification method and device, equipment and a medium, and relates to the technical field of farmland monitoring mapping, and the method comprises the steps: calling a multi-stage feature expression model to process a current element array, so as to obtain a current first temporary storage variable, and setting the plot content as a current second temporary storage variable; updating the current count, judging whether the current count is greater than the quantity of the remote sensing data or not, and if not, skipping to the step of processing the current element array; if yes, determining a land parcel prediction model based on the current first temporary storage variable, the current second temporary storage variable and the initial long short-term memory network; and processing a to-be-processed temporary storage variable obtained by inputting the target element array of the current to-be-identified land parcel into a preset multi-stage growth feature expression model by using the land parcel prediction model to obtain a prediction result so as to determine whether the current to-be-identified land parcel is a rice land parcel or not. Therefore, the accuracy of rice identification on the land parcels can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of farmland monitoring and mapping, and in particular to a rice plot identification method, device, equipment and medium. Background Art

[0002] Radar remote sensing can penetrate clouds and rain and fog to conduct observations in any weather conditions. Radar remote sensing can be used to obtain structural information on rice growth through different polarization methods, such as the height, density and scattering of objects and vegetation on the ground. With a high revisit rate, radar remote sensing data performs well in distinguishing rice fields from other crops. Through the identification and monitoring of large areas of rice fields by radar remote sensing, the government and farmers can better carry out agricultural planning and management, improve agricultural production efficiency and the scientific nature of decision-making.

[0003] At present, the main methods used to identify rice plots using radar remote sensing data are as follows:

[0004] First, the different polarization modes of radar signals are used to analyze the scattering characteristics of ground objects, and the rice is identified based on the range of characteristic values ​​corresponding to a specific period of time;

[0005] Second, construct ground texture features based on radar, and then use neural network, support vector machine and other methods to identify rice;

[0006] 3. Time series analysis is performed based on radar time series data with consecutive numbers over a period of time to distinguish rice from other crops.

[0007] However, in actual work, the above method will encounter the following difficulties:

[0008] First, radar remote sensing data usually has a low resolution. The plot corresponding to a pixel may be mixed with other plots, such as farmers planting corn in rice fields or rice adjacent to water bodies, which makes operations based on feature "values" unstable;

[0009] Second, data at certain time points may not be sufficient to distinguish rice from other ground objects. For example, at the heading stage of rice, the growth states of rice and corn are very similar, making it difficult to distinguish rice from other aquatic plants at the early stage of rice seedling emergence;

[0010] Third, time series data is seriously affected by satellite replay time. In a period of time, the phase difference between different plots can reach more than 10 days, which has a very serious impact on the rice time series change curve. Therefore, the above existing methods only perform well in laboratories and test areas, and when it comes to rice identification in a large area, the accuracy of the above rice identification methods is low.

[0011] As can be seen from the above, how to improve the accuracy of rice identification in a large area during the rice plot identification process is a problem that needs to be solved urgently. Summary of the invention

[0012] In view of this, the purpose of the present invention is to provide a rice plot identification method, device, equipment and medium, which can analyze the radar remote sensing data corresponding to the growth stages of the plot to be identified at different periods, and use the results of the analysis to train the initial plot prediction model to obtain a plot prediction model, and use the plot prediction model to predict whether the plot to be identified is a rice plot. The specific scheme is as follows:

[0013] In a first aspect, the present application provides a rice plot identification method, comprising:

[0014] The current element array in the radar remote sensing data is processed by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain a current first temporary variable, and the plot content corresponding to the current element array is set as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified;

[0015] Update the current count, and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if greater, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network;

[0016] Inputting the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and inputting the temporary variables to be processed into the plot prediction model for processing to obtain prediction results;

[0017] Based on the prediction result, it is determined whether the current plot to be identified is a rice plot.

[0018] Optionally, before the method of calling several stage feature expression models through the preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data, it also includes:

[0019] Obtain radar remote sensing data corresponding to the farmland to be identified and a preset number of remote sensing data;

[0020] Clustering the current element array using a preset clustering rule to obtain a first clustering array, a second clustering array, a third clustering array, and a fourth clustering array with cluster center values ​​from large to small;

[0021] Determine a radar data water surface reference value based on the fourth cluster array, determine a radar data common vegetation reference value based on the second cluster array, determine a radar data dense vegetation reference value based on the first cluster array, and determine a radar basic change reference value based on the first cluster array, the second cluster array, the third cluster array, and the fourth cluster array;

[0022] Among them, the preset number of remote sensing data is the number of data elements in the radar remote sensing data; each data element in the radar remote sensing data includes a plot number, an element array and a plot content, and each data element uniquely corresponds to a plot to be identified.

[0023] Optionally, the calling of several stage feature expression models by a preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data to obtain the current first temporary variable includes:

[0024] By presetting a multi-stage growth feature expression model, a first variable is determined based on a current element array, a water surface reference value of the radar data, and a dense vegetation reference value of the radar data;

[0025] The first-stage feature expression model is called by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model, and a first target array is determined based on the first-stage sub-array in the current element array to store the first target array in the second variable;

[0026] The second-stage feature expression model is called by the preset multi-stage growth feature expression model to utilize the second-stage feature expression model, and a second target array is determined based on the second-stage sub-array in the current element array to store the second target array in a third variable;

[0027] The third stage feature expression model is called by the preset multi-stage growth feature expression model to utilize the third stage feature expression model, and a third target array is determined based on the third stage sub-array in the current element array to store the third target array in a fourth variable;

[0028] The first variable, the second variable, the third variable and the fourth variable are stored in the current first temporary variable;

[0029] Among them, the first stage is the growth stage from the seedling stage to the tillering stage of rice; the second stage is the growth stage from the jointing stage to the heading stage of rice; and the third stage is the growth stage from the flowering stage to the harvesting stage of rice.

[0030] Optionally, calling the first-stage feature expression model by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model and determining the first target array based on the first-stage sub-array in the current element array includes:

[0031] The first stage feature expression model is called by presetting the multi-stage growth feature expression model, so as to utilize the first stage feature expression model, and based on the radar data water surface reference value, the radar basic change reference value, the radar data common vegetation reference value, the radar data dense vegetation reference value and the corresponding time series data in the current element array, respectively determine the rice seedling emergence water injection characteristics, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, the growth characteristics of the tillering stage and the normalized result of the interval from the vegetation to the water surface;

[0032] The water injection characteristics at the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, and the growth characteristics at the tillering stage are used as the first target array of the output of the feature expression model of the first stage.

[0033] Optionally, calling the second-stage feature expression model by presetting the multi-stage growth feature expression model to utilize the second-stage feature expression model, and determining the second target array based on the second-stage sub-array in the current element array to store the second target array in the third variable includes:

[0034] The second-stage feature expression model is called by presetting the multi-stage growth feature expression model, so as to utilize the second-stage feature expression model, and based on the common vegetation reference value of the radar data, the basic change reference value of the radar, and the corresponding time series data in the current element array, respectively determine the growth characteristics of rice in the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalization result based on the vegetation density;

[0035] The growth characteristics of the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalized results based on vegetation density are used as the second target array of the output of the second stage feature expression model.

[0036] Optionally, calling the third stage feature expression model by presetting the multi-stage growth feature expression model to utilize the third stage feature expression model, and determining a third target array based on the third stage sub-array in the current element array to store the third target array in a fourth variable includes:

[0037] The third-stage feature expression model is called by the preset multi-stage growth feature expression model, so as to utilize the third-stage feature expression model, and determine the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvest feature and the normalized result of the vegetation density of rice respectively based on the dense vegetation reference value of the radar data, the common vegetation reference value of the radar data and the corresponding time series data in the current element array;

[0038] The first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvesting feature and the normalized result of the vegetation density are used as a third target array of the output of the third stage feature expression model.

[0039] Optionally, determining the land parcel prediction model based on the current first temporarily stored variable, the current second temporarily stored variable, and the initial long short-term memory network includes:

[0040] Inputting the current first temporary stored variable into the initial long short-term memory network to obtain an initial prediction result output by the initial long short-term memory network, and comparing the current second temporary stored variable with the initial prediction result to obtain a comparison result;

[0041] If the comparison result indicates that the initial long short-term memory network fails to predict, the model parameters corresponding to the initial long short-term memory network are adjusted based on the comparison result.

[0042] In a second aspect, the present application provides a rice plot identification device, comprising:

[0043] A temporary variable determination module is used to process the current element array in the radar remote sensing data by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain the current first temporary variable, and set the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified;

[0044] The model acquisition module is used to update the current count and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if greater, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network;

[0045] A prediction result determination module is used to input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain prediction results;

[0046] The prediction result judgment module is used to determine whether the current plot to be identified is a rice plot based on the prediction result.

[0047] In a third aspect, the present application provides an electronic device, including:

[0048] Memory, used to store computer programs;

[0049] A processor is used to execute the computer program to implement the aforementioned rice plot identification method.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program implements the aforementioned rice plot identification method when executed by a processor.

[0051] As can be seen from the above, before the present application identifies the rice plots, it is necessary to process the current element array in the radar remote sensing data by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain the current first temporary variable, and set the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified; update the current count, and then determine whether the current count is greater than the preset count. Assume the number of remote sensing data, if it is not greater than, then jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if it is greater than, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network; input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain the temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain the prediction result; determine whether the current plot to be identified is a rice plot based on the prediction result. It can be seen that the present application analyzes the radar remote sensing data corresponding to the growth stages of the plot to be identified at different periods, and uses the results of the analysis to train the initial plot prediction model to obtain the plot prediction model, and uses the plot prediction model to predict whether the plot to be identified is a rice plot. In this way, the accuracy of rice identification in plots in a large area is improved, and the user experience is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0053] Figure 1 A flow chart of a rice plot identification method disclosed in this application;

[0054] Figure 2 This is a schematic diagram of the structure of a rice plot identification device disclosed in this application;

[0055] Figure 3 This is a structural diagram of an electronic device disclosed in this application. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] Radar remote sensing can be used to obtain structural information of rice growth through different polarization modes, such as the height, density and scattering of plot objects and vegetation; currently, the main means of using radar remote sensing data to identify rice plots only performs well in laboratories and experimental areas, and when identifying rice in a large area, the accuracy of the above rice identification method is low. To this end, the present application provides a rice plot identification method, which can analyze the radar remote sensing data corresponding to the growth stages of the plot to be identified at different periods, and use the results of the analysis to train the initial plot prediction model to obtain a plot prediction model, and use the plot prediction model to predict whether the plot to be identified is a rice plot, thereby improving the accuracy of rice identification in plots in a large area.

[0058] See also Figure 1 As shown, the embodiment of the present invention discloses a rice plot identification method, comprising:

[0059] Step S11, calling several stage feature expression models through a preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data to obtain the current first temporary variable, and setting the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified.

[0060] It is worth mentioning that the growth of rice is different from that of other crops in that it can show different physical characteristics at different growth stages, such as a relatively smooth surface at the beginning of the emergence period, a dense vegetation surface before harvest, and a smooth surface of the newly raised land after harvest. It is worth mentioning that the above physical characteristics are an effective supplement to the key characteristics of rice and its time series characteristics.

[0061] In this embodiment, it is necessary to obtain radar data information LDData of several plots corresponding to farmland in a region, the number of radar data information LDNumber, radar data water surface reference value LDSM, radar data ordinary vegetation reference value LDPT, radar data dense vegetation reference value LDMJ and radar base change reference value LDBH. Among them, the number of radar data information LDNumber is consistent with the number of elements in the radar data information LDData. Specifically, before the preset multi-stage growth feature expression model is used to call several stage feature expression models to process the current element array in the radar remote sensing data, it can also include: obtaining radar remote sensing data corresponding to the farmland to be identified and a preset number of remote sensing data; clustering the current element array using a preset clustering rule to obtain a first clustering array, a second clustering array, a third clustering array, and a fourth clustering array with cluster center values ​​from large to small; determining a radar data water surface reference value based on the fourth clustering array, determining a radar data common vegetation reference value based on the second clustering array, determining a radar data dense vegetation reference value based on the first clustering array, and determining a radar basic change reference value based on the first clustering array, the second clustering array, the third clustering array, and the fourth clustering array; wherein the preset number of remote sensing data is the number of data elements in the radar remote sensing data; each data element in the radar remote sensing data includes a plot number, an element array, and a plot content, and each data element uniquely corresponds to a plot to be identified.

[0062] It is worth mentioning that the radar data information LDData is a list of data. Each element in the list corresponds to a plot to be identified, and each element contains the following fields: LDDKID, LDRadarArray, and LDCategory. Among them, LDDKID is the plot number, its type is an integer, and the plot number uniquely identifies a plot to be identified; LDRadarArray is a floating-point array of 10 elements, in which each element in the array represents the VH value (Vertical Transmission & HorizontalReception, i.e., the polarization mode of vertical transmission and horizontal reception), in the above floating-point array, the first element is the minimum value of the radar remote sensing data corresponding to the plot during the period from the 1st day to the 12th day of rice sowing; it can be understood that the nth element is the minimum value of the radar remote sensing data corresponding to the plot during the period from the (n-1)*12+1st day to the (n-1)*12+12th day of rice sowing; the VH value is the value of the radar remote sensing data transmitting horizontally polarized electromagnetic waves and receiving vertically polarized electromagnetic waves; LDCategory is the content of the above plot, when LDCategory is 1, it means that the crop planted in the plot is rice, and when it is 0, it means that the crop in the plot is not rice.

[0063] Among them, the radar data reference expression array LDCZBD is an array formed by combining the LDRadarArray fields corresponding to all elements in the radar data information, that is, at this time, the radar data reference expression array is a one-dimensional array of 10 elements of the number of radar data information LDNumber.

[0064] It is worth mentioning that after obtaining the radar data reference expression array, the radar data reference expression array needs to be clustered by K-means. In a specific implementation, the number of clusters for the radar data reference expression array is 4, and in descending order of cluster center values, they are: radar data reference expression array first cluster LDCZBD1, radar data reference expression array second cluster LDCZBD2, radar data reference expression array third cluster LDCZBD3, radar data reference expression array fourth cluster LDCZBD4.

[0065] In a first specific implementation manner, the radar data surface reference value LDSM is a value determined based on the fourth cluster LDCZBD4 of the radar data reference expression array, and the expression formula is as follows:

[0066] The radar data surface reference value LDSM = the cluster center point of LDCZBD4 - 2 × std (LDCZBD4); where std is the calculated standard deviation;

[0067] In a second specific implementation, the radar data common vegetation reference value LDPT is a value determined based on the second cluster LDCZBD2 of the radar data reference expression array, and the expression formula is as follows:

[0068] The cluster center point of the common vegetation reference value LDPT=LDCZBD2 of the radar data;

[0069] In a third specific implementation, the radar data dense vegetation reference value LDMJ is a value determined based on the first cluster LDCZBD1 of the radar data reference expression array, and the expression formula is as follows:

[0070] The reference value of dense vegetation in radar data LDMJ = the cluster center point of LDCZBD1 + 2 × std (LDCZBD1);

[0071] In a fourth specific implementation, the radar basic change reference value LDBH is a value determined based on the radar data reference expression array first cluster LDCZBD1, the radar data reference expression array second cluster LDCZBD2, the radar data reference expression array third cluster LDCZBD3, and the radar data reference expression array fourth cluster LDCZBD4, and the expression formula is as follows:

[0072] Radar basic change reference value LDBH=(std(LDCZBD4)+std(LDCZBD3)+std(LDCZBD2)+std(LDCZBD1)) / 4.

[0073] In this embodiment, it is necessary to construct a radar feature expression module Mcmfp for expressing features of radar data corresponding to rice in the seedling stage to the tillering stage. Among them, the input of the radar feature expression module Mcmfp is an array variable McmfpInput of 10 elements, and the output is an array variable McmfpOutput of 4 elements. Specifically, the method of calling several stage feature expression models through a preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data to obtain the current first temporary variable may include: determining the first variable through a preset multi-stage growth feature expression model and based on the current element array, the water surface reference value of the radar data, and the dense vegetation reference value of the radar data; calling the first stage feature expression model through a preset multi-stage growth feature expression model to utilize the first stage feature expression model and determine the first target array based on the first stage sub-array in the current element array to store the first target array in the second variable; calling the second stage feature expression model through a preset multi-stage growth feature expression model to utilize the first stage sub-array in the current element array to obtain the first temporary variable. A two-stage feature expression model is provided, and a second target array is determined based on the second stage sub-array in the current element array, so as to store the second target array in a third variable; a third stage feature expression model is called by a preset multi-stage growth feature expression model, so as to utilize the third stage feature expression model, and a third target array is determined based on the third stage sub-array in the current element array, so as to store the third target array in a fourth variable; the first variable, the second variable, the third variable and the fourth variable are stored in the current first temporary variable; wherein the first stage is the growth stage from the seedling stage to the tillering stage of rice; the second stage is the growth stage from the jointing stage to the booting stage of rice; and the third stage is the growth stage from the flowering stage to the harvesting stage of rice.

[0074] In a specific implementation, the embodiment of the present application may establish a floating-point array of 4 elements whose initial values ​​are all 0, so as to subsequently save the output results of the radar feature expression module Mcmfp.

[0075] In this embodiment, when the radar feature expression module Mcmfp processes the array variable McmfpInput of 10 elements, the 4 elements output are determined based on the water injection characteristics of the rice field in the seedling stage, the rice growth characteristics at the end of the seedling stage, the difference characteristics of rice between the tillering stage and the beginning of the seedling stage, and the rice growth characteristics in the tillering stage.

[0076] Among them, the expression formula of water injection characteristics of rice fields in the seedling stage is as follows:

[0077] McmfpOutput[1]=(McmfpInput[1]-LDSM) / (McmfpInput[2]-LDSM);

[0078] The expression formula of rice growth characteristics at the end of the emergence period is as follows:

[0079] McmfpOutput[2]=(McmfpInput[2]-McmfpInput[3])-LDBH;

[0080] The expression formula for the difference characteristics of rice at the tillering stage and the early emergence stage is as follows:

[0081] McmfpOutput[3]=(McmfpInput[3]-McmfpInput[1]);

[0082] The expression formula of rice growth characteristics in the tillering stage is as follows:

[0083] McmfpOutput[4]=(McmfpInput[5]-LDPT);

[0084] Subsequently, the water injection characteristics of the rice field at the seedling stage, the rice growth characteristics at the end of the seedling stage, the difference characteristics of rice between the tillering stage and the beginning of the seedling stage, and the rice growth characteristics at the tillering stage were normalized from the vegetation to the water surface. The normalization formula is as follows:

[0085] McmfpOutput=McmfpOutput / (LDMJ-LDSM);

[0086] Finally, McmfpOutput is output as the result of Mcmfp.

[0087] In this embodiment, a radar feature expression module Mbjys is constructed for expressing the features of the radar data corresponding to the rice from the jointing stage to the booting stage. The input of the radar feature expression module Mbjys is an array variable MbjysInput of 10 elements, and the output is an array variable MbjysOutput of 3 elements. Specifically, calling the first-stage feature expression model by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model, and determining the first target array based on the first-stage sub-array in the current element array, may include: calling the first-stage feature expression model by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model, and determining the water injection characteristics of rice at the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, the growth characteristics at the tillering stage, and the normalized results of the interval from vegetation to the water surface based on the radar data water surface reference value, the radar basic change reference value, the radar data ordinary vegetation reference value, the radar data dense vegetation reference value, and the corresponding time series data in the current element array; and using the water injection characteristics at the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, and the growth characteristics at the tillering stage as the first target array output by the first-stage feature expression model.

[0088] In a specific implementation, the embodiment of the present application may establish an array variable MbjysOutput with 3 elements all being 0, so as to subsequently save the output result of the radar feature expression module Mbjys.

[0089] In this embodiment, when the radar feature expression module Mbjys processes the array variable McmfpInput of 10 elements, the three elements output are determined based on the growth characteristics of rice in the jointing stage, the growth characteristics of rice in the booting stage, and the growth characteristics of rice in the booting stage compared with the jointing stage.

[0090] Among them, the expression formula of rice growth characteristics in the jointing stage is as follows:

[0091] MbjysOutput[1]=MbjysInput[5]-(LDPT-LDBH);

[0092] The expression formula of rice growth characteristics at the booting stage is as follows:

[0093] MbjysOutput[2]=MbjysInput[6]-(LDPT-LDBH);

[0094] The expression formula of rice growth characteristics at booting stage compared with jointing stage is as follows:

[0095] MbjysOutput[3]=MbjysInput[6]-MbjysInput[5];

[0096] Subsequently, the rice growth characteristics at the jointing stage, the rice growth characteristics at the booting stage, and the rice growth characteristics at the booting stage compared with the jointing stage were normalized based on vegetation density, wherein the normalization formula is as follows:

[0097] MbjysOutput=MbjysOutput / (LDMJ-LDPT);

[0098] Finally, MbjysOutput is output as the result of the radar feature expression module Mbjys.

[0099] In this embodiment, a radar feature expression module Mkhsg is constructed for expressing the features of radar data corresponding to the period from the flowering stage of rice to the completion of harvesting, wherein the input of Mkhsg is an array variable MkhsgInput of 10 elements, and the output is an array variable MkhsgOutput of 4 elements. Specifically, calling the second-stage feature expression model through the preset multi-stage growth feature expression model to utilize the second-stage feature expression model, and determining the second target array based on the second-stage sub-array in the current element array to store the second target array in the third variable may include: calling the second-stage feature expression model through the preset multi-stage growth feature expression model to utilize the second-stage feature expression model, and determining the growth characteristics of rice in the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalized results based on vegetation density based on the common vegetation reference value of the radar data, the radar basic change reference value, and the corresponding time series data in the current element array; using the growth characteristics of the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalized results based on vegetation density as the second target array of the output of the second-stage feature expression model.

[0100] In a specific implementation, the embodiment of the present application may establish an array variable MkhsgOutput of three elements all of which are 0, so as to subsequently save the output result of the radar feature expression module Mkhsg.

[0101] In this embodiment, when the radar feature expression module Mkhsg processes the input array variable McmfpInput of 10 elements, the 4 elements output are determined based on the first feature of persistent dense vegetation, the second feature of persistent dense vegetation, the rice growth tacit vegetation feature and the rice harvesting feature. Specifically, calling the third-stage feature expression model through a preset multi-stage growth feature expression model to utilize the third-stage feature expression model, and determining the third target array based on the third-stage sub-array in the current element array to store the third target array in the fourth variable may include: calling the third-stage feature expression model through a preset multi-stage growth feature expression model to utilize the third-stage feature expression model, and determining the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvesting feature and the normalized result of vegetation density of rice based on the dense vegetation reference value of the radar data, the ordinary vegetation reference value of the radar data and the corresponding time series data in the current element array; using the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvesting feature and the normalized result of vegetation density as the third target array output by the third-stage feature expression model.

[0102] Among them, the expression formula of the first dense vegetation feature is as follows:

[0103] MkhsgOutput[1]=MkhsgInput[7]-LDMJ;

[0104] The expression formula of the second dense vegetation feature is as follows:

[0105] MkhsgOutput[2]=MkhsgInput[8]-LDMJ;

[0106] The expression formula of rice growth tacit vegetation characteristics is as follows:

[0107] MkhsgOutput[3]=abs(MkhsgInput[8]-MkhsgInput[7])+abs(MkhsgInput[9]-MkhsgInput[8]);

[0108] The expression formula of rice harvesting characteristics is as follows:

[0109] MkhsgOutput[4]=LDPT-MkhsgInput

[10] ;

[0110] Subsequently, the first feature of persistent dense vegetation, the second feature of persistent dense vegetation, the rice growth tacit vegetation feature, and the rice harvesting feature are normalized based on vegetation density, where the normalization formula is as follows:

[0111] MkhsgOutput=MkhsgOutput / (LDMJ-LDPT);

[0112] Finally, MkhsgOutput is output as the result of the radar feature expression module Mkhsg.

[0113] Step S12, update the current count, and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of calling several stage feature expression models through the preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data; if greater, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network.

[0114] In this embodiment, a multi-stage growth feature expression module Mexpress needs to be constructed so that the multi-stage growth feature expression module Mexpress can perform radar feature expression of rice growth on the input data by calling radar feature expression module Mcmfp, radar feature expression module Mbjys and radar feature expression module Mkhsg. Among them, the input of the multi-stage growth feature expression module Mexpress is an array variable MexpressInput of 10 elements, and the output is an array variable MexpressOutput of rice growth radar features of 21 elements.

[0115] It is worth mentioning that in the multi-stage growth feature expression module Mexpress, it is necessary to first establish a rice growth radar feature array variable MexpressOutput that can store 21 elements. The initial content is all 0. Then the first temporary variable METemp1 is determined, and the determination formula is as follows:

[0116] METemp1 = (MexpressInput-LDSM) / (LDMJ-LDSM);

[0117] Secondly, search for all elements in the first temporary variable METemp1 of the rice growth radar feature, and determine the size relationship between the above elements and the numbers 1 and 0. If an element is greater than 1, change the element to 1; if an element is less than 0, change the element to 0.

[0118] Then, the radar feature expression module Mcmfp is called, and the first temporary variable METemp1 is input into the radar feature expression module Mcmfp to obtain the output result McmfpOutput of the radar feature expression module Mcmfp, and the result McmfpOutput output by the radar feature expression module Mcmfp is stored in the second temporary variable METemp2.

[0119] Furthermore, the radar feature expression module Mbjys is called, and the first temporary variable METemp1 is input into the radar feature expression module Mbjys to obtain the output result MbjysOutput of the radar feature expression module Mbjys, and the output result MbjysOutput of the radar feature expression module Mbjys is stored in the third temporary variable METemp3.

[0120] Subsequently, the radar feature expression module Mkhsg is called, and the first temporary variable METemp1 is input into the radar feature expression module Mkhsg to obtain the result MkhsgOutput output by the radar feature expression module Mkhsg, and the result MkhsgOutput output by the radar feature expression module Mkhsg is stored in the fourth temporary variable METemp4.

[0121] Finally, the first temporary variable METemp1, the second temporary variable METemp2, the third temporary variable METemp3 and the fourth temporary variable METemp4 are combined into an array and stored in the rice growth radar feature array variable MexpressOutput, and the rice growth radar feature array variable MexpressOutput is used as the output result of the multi-stage growth feature expression module Mexpress.

[0122] Step S13, input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain prediction results.

[0123] In this embodiment, the radar data information LDData is processed using the multi-stage growth feature expression module Mexpress to obtain an attribute table LDTable including rice radar time series and growth characteristics, and the data in the attribute table LDTable is used as input data to train the long short-term memory network model LSTM.

[0124] It is worth mentioning that the attribute table LDTable contains two fields: LDSequence and LDDecsion, where LDSequence is an array of 21 elements, and the elements in the array describe the radar timing and growth characteristics of rice; LDDecsion is an integer variable. If LDDecsion is 1, it means that the crop planted in the plot is rice. If LDDecsion is 0, it means that the crop planted in the plot is not rice.

[0125] Subsequently, a counter TableCounter is created for the attribute table, and the initial value of the counter TableCounter is set to 1.

[0126] Then, the multi-stage growth feature expression module Mexpress is used to process the LDRadarArray in the plot corresponding to the value of the counter TableCounter in the radar data information LDData to obtain the output result MexpressOutput of Mexpress, and MexpressOutput is stored in the current first temporary variable LDTTemp1. Accordingly, the embodiment of the present application needs to use the multi-stage growth feature expression module Mexpress to extract the LDDecsion field in the plot corresponding to the value of the counter TableCounter in the radar data information LDData, and store the extracted LDDecsion field to establish the current second temporary variable LDTTemp2.

[0127] Subsequently, a row of data is created, and then the LDSequence field of the row of data is set to the current first temporary variable LDTTemp1, and the LDDecsion field of the row of data is set to the current second temporary variable LDTTemp2, and the current first temporary variable and the current second temporary variable are stored in the attribute table LDTable.

[0128] After the current first temporary variable and the current second temporary variable are stored in the attribute table LDTable, the counter TableCounter corresponding to the attribute table needs to be incremented by one, and the expression is as follows:

[0129] TableCounter=TableCounter+1;

[0130] Furthermore, determine whether the counter TableCounter is greater than the number of radar data information LDNumber. If the counter is not greater than the number of radar data information, jump to the step of using the multi-stage growth feature expression module Mexpress to process the LDRadarArray in the plot corresponding to the value of the counter TableCounter in the radar data information LDData; if the counter is greater than the number of radar data information, perform the following operations.

[0131] In this embodiment, an initial long short-term memory network model LSTM is established, wherein the input of the initial long short-term memory network model is time series data including 21 elements, and the output is 1 regression prediction value. Subsequently, the LDSequence field in the attribute table LDTable is input into the initial long short-term memory network model LSTM, so that the initial long short-term memory network model outputs a regression prediction value, wherein the LDDecsion field in the attribute table LDTable is used to compare with the above regression prediction value, so as to adjust the model parameters of the initial long short-term memory network model based on the comparison result, thereby improving the performance of the long short-term memory network model.

[0132] In this embodiment, the radar remote sensing data Test of the land to be identified needs to be input into the multi-stage growth feature expression module Mexpress, so that Mexpress can be used to express the features of the radar remote sensing data Test, and then the results output by the multi-stage growth feature expression module Mexpress are used to train the long short-term memory network model LSTM, so that the long short-term memory network model can be used to judge whether the land to be identified belongs to a rice field.

[0133] It is worth mentioning that the embodiment of the present application inputs the radar remote sensing data Test corresponding to the land parcel to be identified into the multi-stage growth feature expression module Mexpress, wherein the structure of Test is the same as the LDRadarArray structure in LDData.

[0134] Subsequently, the radar remote sensing data Test of the land parcel to be identified is input into the multi-stage growth feature expression module Mexpress for processing, and the result MexpressOutput output by the multi-stage growth feature expression module Mexpress is stored in the first temporary variable RCTemp1.

[0135] Then, the first temporary variable RCTemp1 is processed using the long short-term memory network model LSTM, and the result output by the long short-term memory network model LSTM is stored in the second temporary variable RCTemp2, wherein the second temporary variable RCTemp2 is the regression prediction value. Specifically, the method of determining the land prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network may include: inputting the current first temporary variable into the initial long short-term memory network to obtain the initial prediction result output by the initial long short-term memory network, and comparing the current second temporary variable with the initial prediction result to obtain a comparison result; if the comparison result indicates that the initial long short-term memory network prediction fails, adjusting the model parameters corresponding to the initial long short-term memory network based on the comparison result.

[0136] Step S14: determining whether the current plot to be identified is a rice plot based on the prediction result.

[0137] In a specific embodiment, after obtaining the second temporary variable RCTemp2, the embodiment of the present application needs to determine whether the second temporary variable RCTemp2 is greater than a preset threshold; if the second temporary variable RCTemp2 is not greater than the preset threshold, the plot to be identified is determined to be a non-rice plot; if the second temporary variable RCTemp2 is greater than the preset threshold, the plot to be identified is determined to be a rice plot.

[0138] As can be seen from the above, before the rice plot identification is performed, the embodiment of the present application needs to analyze the radar remote sensing data corresponding to the growth stages of the plot to be identified at different periods, and use the results of the analysis to train the initial plot prediction model to obtain a plot prediction model, and use the plot prediction model to predict whether the plot to be identified is a rice plot. In this way, the accuracy of rice identification in plots in a large area is improved, and the user experience is enhanced.

[0139] Accordingly, see Figure 2 As shown, the present application also provides a rice plot identification device, comprising:

[0140] The temporary variable determination module 11 is used to process the current element array in the radar remote sensing data by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain the current first temporary variable, and set the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified;

[0141] The model acquisition module 12 is used to update the current count and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if greater, determine the land parcel prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network;

[0142] The prediction result determination module 13 is used to input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain prediction results;

[0143] The prediction result judgment module 14 is used to determine whether the current plot to be identified is a rice plot based on the prediction result.

[0144] As can be seen from the above, before the rice plot identification is performed, the embodiment of the present application needs to process the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model to obtain the current first temporary variable, and set the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified; update the current count, and then determine whether the current count is greater than The number of remote sensing data is preset. If it is not greater than, jump to the step of calling several stage feature expression models through the preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data; if it is greater than, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network; input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain the temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain the prediction result; determine whether the current plot to be identified is a rice plot based on the prediction result. In this way, the accuracy of rice identification in plots in a large area is improved, and the user experience is improved.

[0145] In some specific embodiments, the rice plot identification device may further include:

[0146] A data acquisition unit, used to acquire radar remote sensing data corresponding to the farmland to be identified and a preset number of remote sensing data;

[0147] An array clustering unit, used for clustering the current element array using a preset clustering rule to obtain a first clustering array, a second clustering array, a third clustering array and a fourth clustering array with cluster center values ​​from large to small;

[0148] A reference value determination unit is used to determine a water surface reference value of radar data based on the fourth cluster array, determine a common vegetation reference value of radar data based on the second cluster array, determine a dense vegetation reference value of radar data based on the first cluster array, and determine a radar basic change reference value based on the first cluster array, the second cluster array, the third cluster array, and the fourth cluster array; wherein the preset number of remote sensing data is the number of data elements in the radar remote sensing data; each data element in the radar remote sensing data includes a plot number, an element array, and a plot content, and each data element uniquely corresponds to a plot to be identified.

[0149] In some specific embodiments, the rice plot identification device may further include:

[0150] A first variable determination unit, configured to determine a first variable by presetting a multi-stage growth characteristic expression model and based on a current element array, a water surface reference value of the radar data, and a dense vegetation reference value of the radar data;

[0151] A second variable determination unit is used to call the first-stage feature expression model by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model, and determine a first target array based on the first-stage sub-array in the current element array to store the first target array in the second variable;

[0152] a third variable determination unit, configured to call the second-stage feature expression model by a preset multi-stage growth feature expression model, so as to utilize the second-stage feature expression model, and determine a second target array based on the second-stage sub-array in the current element array, so as to store the second target array in the third variable;

[0153] a fourth variable determination unit, configured to call the third stage feature expression model by a preset multi-stage growth feature expression model, to utilize the third stage feature expression model, and to determine a third target array based on the third stage sub-array in the current element array, to store the third target array in a fourth variable;

[0154] The temporary variable determination unit is used to store the first variable, the second variable, the third variable and the fourth variable into the current first temporary variable; wherein the first stage is the growth stage from the seedling stage to the tillering stage of rice; the second stage is the growth stage from the jointing stage to the booting stage of rice; and the third stage is the growth stage from the flowering stage to the harvesting stage of rice.

[0155] In some specific embodiments, the rice plot identification device may further include:

[0156] A first model calling unit is used to call the first stage feature expression model through a preset multi-stage growth feature expression model, so as to use the first stage feature expression model and determine the water injection characteristics of rice at the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, the growth characteristics at the tillering stage, and the normalized result of the interval from the vegetation to the water surface based on the radar data water surface reference value, the radar basic change reference value, the radar data ordinary vegetation reference value, the radar data dense vegetation reference value, and the corresponding time series data in the current element array;

[0157] The first array determination unit is used to use the water injection characteristics of the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, and the growth characteristics of the tillering stage as the first target array of the output of the first stage feature expression model.

[0158] In some specific embodiments, the rice plot identification device may further include:

[0159] A second model calling unit is used to call the second stage feature expression model by presetting the multi-stage growth feature expression model, so as to use the second stage feature expression model, and determine the growth characteristics of rice in the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalization result based on the vegetation density based on the common vegetation reference value of the radar data, the radar basic change reference value, and the corresponding time series data in the current element array;

[0160] The second array determination unit is used to use the growth characteristics of the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalized results based on vegetation density as the second target array of the output of the second stage feature expression model.

[0161] In some specific embodiments, the rice plot identification device may further include:

[0162] A third model calling unit is used to call the third stage feature expression model by a preset multi-stage growth feature expression model, so as to use the third stage feature expression model and determine the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvest feature and the normalized result of the vegetation density of rice based on the dense vegetation reference value of the radar data, the common vegetation reference value of the radar data and the corresponding time series data in the current element array;

[0163] The third array determination unit is used to use the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvesting feature and the normalized result of the vegetation density as the third target array of the output of the third stage feature expression model.

[0164] In some specific implementations, the model acquisition module 12 may specifically include:

[0165] A comparison result determination unit, used for inputting the current first temporary stored variable into the initial long short-term memory network to obtain an initial prediction result output by the initial long short-term memory network, and comparing the current second temporary stored variable with the initial prediction result to obtain a comparison result;

[0166] A parameter adjustment unit is used to adjust the model parameters corresponding to the initial long short-term memory network based on the comparison result if the comparison result indicates that the initial long short-term memory network fails to predict.

[0167] Furthermore, the present application also discloses an electronic device. Figure 3 It is a structural diagram of an electronic device 20 shown according to an exemplary embodiment, and the content in the figure cannot be regarded as any limitation on the scope of use of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. Among them, the memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to implement the relevant steps in the rice plot identification method disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0168] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0169] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0170] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, and can be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program that can be used to complete the rice plot identification method performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include computer programs that can be used to complete other specific tasks.

[0171] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein the computer program, when executed by a processor, implements the rice plot identification method disclosed above. The specific steps of the method can refer to the corresponding contents disclosed in the above embodiments, and will not be repeated here.

[0172] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0173] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0174] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0175] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0176] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A rice plot identification method, characterized in that: include: The current element array in the radar remote sensing data is processed by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain a current first temporary variable, and the plot content corresponding to the current element array is set as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified; Update the current count, and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if greater, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network; Inputting the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and inputting the temporary variables to be processed into the plot prediction model for processing to obtain prediction results; Based on the prediction result, it is determined whether the current plot to be identified is a rice plot.

2. The rice plot identification method according to claim 1, characterized in that: Before the method of calling several stage feature expression models by the preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data, it also includes: Obtain radar remote sensing data corresponding to the farmland to be identified and a preset number of remote sensing data; Clustering the current element array using a preset clustering rule to obtain a first clustering array, a second clustering array, a third clustering array, and a fourth clustering array with cluster center values ​​from large to small; Determine a radar data water surface reference value based on the fourth cluster array, determine a radar data common vegetation reference value based on the second cluster array, determine a radar data dense vegetation reference value based on the first cluster array, and determine a radar basic change reference value based on the first cluster array, the second cluster array, the third cluster array, and the fourth cluster array; Among them, the preset number of remote sensing data is the number of data elements in the radar remote sensing data; each data element in the radar remote sensing data includes a plot number, an element array and a plot content, and each data element uniquely corresponds to a plot to be identified.

3. The rice plot identification method according to claim 2, characterized in that: The method of calling several stage feature expression models through the preset multi-stage growth feature expression model to process the current element array in the radar remote sensing data to obtain the current first temporary variable includes: By presetting a multi-stage growth feature expression model, a first variable is determined based on a current element array, a water surface reference value of the radar data, and a dense vegetation reference value of the radar data; The first-stage feature expression model is called by presetting the multi-stage growth feature expression model to utilize the first-stage feature expression model, and a first target array is determined based on the first-stage sub-array in the current element array to store the first target array in the second variable; The second-stage feature expression model is called by the preset multi-stage growth feature expression model to utilize the second-stage feature expression model, and a second target array is determined based on the second-stage sub-array in the current element array to store the second target array in a third variable; The third stage feature expression model is called by the preset multi-stage growth feature expression model to utilize the third stage feature expression model, and a third target array is determined based on the third stage sub-array in the current element array to store the third target array in a fourth variable; The first variable, the second variable, the third variable and the fourth variable are stored in the current first temporary variable; Among them, the first stage is the growth stage from the seedling stage to the tillering stage of rice; the second stage is the growth stage from the jointing stage to the heading stage of rice; and the third stage is the growth stage from the flowering stage to the harvesting stage of rice.

4. The rice plot identification method according to claim 3, characterized in that: The calling of the first-stage feature expression model by the preset multi-stage growth feature expression model to utilize the first-stage feature expression model and determine the first target array based on the first-stage sub-array in the current element array includes: The first stage feature expression model is called by presetting the multi-stage growth feature expression model, so as to utilize the first stage feature expression model, and based on the radar data water surface reference value, the radar basic change reference value, the radar data common vegetation reference value, the radar data dense vegetation reference value and the corresponding time series data in the current element array, respectively determine the rice seedling emergence water injection characteristics, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, the growth characteristics of the tillering stage and the normalized result of the interval from the vegetation to the water surface; The water injection characteristics at the seedling stage, the growth characteristics at the end of the seedling stage, the difference characteristics between the tillering stage and the beginning of the seedling stage, and the growth characteristics at the tillering stage are used as the first target array of the output of the feature expression model of the first stage.

5. The rice plot identification method according to claim 3, characterized in that: The calling of the second stage feature expression model by the preset multi-stage growth feature expression model to utilize the second stage feature expression model and determining the second target array based on the second stage sub-array in the current element array to store the second target array in the third variable includes: The second-stage feature expression model is called by presetting the multi-stage growth feature expression model, so as to utilize the second-stage feature expression model, and based on the common vegetation reference value of the radar data, the basic change reference value of the radar, and the corresponding time series data in the current element array, respectively determine the growth characteristics of rice in the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalization result based on the vegetation density; The growth characteristics of the jointing stage, the growth characteristics of the booting stage, the growth characteristics of the booting stage compared with the jointing stage, and the normalized results based on vegetation density are used as the second target array of the output of the second stage feature expression model.

6. The rice plot identification method according to claim 3, characterized in that: The calling of the third stage feature expression model by the preset multi-stage growth feature expression model to utilize the third stage feature expression model and determining a third target array based on the third stage sub-array in the current element array to store the third target array in a fourth variable includes: The third-stage feature expression model is called by the preset multi-stage growth feature expression model, so as to utilize the third-stage feature expression model, and determine the first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvest feature and the normalized result of the vegetation density of rice respectively based on the dense vegetation reference value of the radar data, the common vegetation reference value of the radar data and the corresponding time series data in the current element array; The first dense vegetation feature, the second dense vegetation feature, the growth tacit vegetation feature, the harvesting feature and the normalized result of the vegetation density are used as a third target array of the output of the third stage feature expression model.

7. The rice plot identification method according to claim 1, characterized in that: The determining of the land parcel prediction model based on the current first temporary stored variable, the current second temporary stored variable and the initial long short-term memory network includes: Inputting the current first temporary stored variable into the initial long short-term memory network to obtain an initial prediction result output by the initial long short-term memory network, and comparing the current second temporary stored variable with the initial prediction result to obtain a comparison result; If the comparison result indicates that the initial long short-term memory network fails to predict, the model parameters corresponding to the initial long short-term memory network are adjusted based on the comparison result.

8. A rice plot identification device, characterized in that: include: A temporary variable determination module is used to process the current element array in the radar remote sensing data by calling several stage feature expression models through a preset multi-stage growth feature expression model to obtain the current first temporary variable, and set the plot content corresponding to the current element array as the current second temporary variable; wherein the current element array is an element array storing time series data corresponding to the current count, and each element is arranged in order from small to large according to the number of rice sowing days; the radar remote sensing data includes the plot number, element array and plot content corresponding to each plot to be identified; The model acquisition module is used to update the current count and then determine whether the current count is greater than the preset number of remote sensing data. If not, jump to the step of processing the current element array in the radar remote sensing data by calling several stage feature expression models through the preset multi-stage growth feature expression model; if greater, determine the plot prediction model based on the current first temporary variable, the current second temporary variable and the initial long short-term memory network; A prediction result determination module is used to input the target element array corresponding to the current plot to be identified into the preset multi-stage growth feature expression model to obtain temporary variables to be processed, and input the temporary variables to be processed into the plot prediction model for processing to obtain prediction results; The prediction result judgment module is used to determine whether the current plot to be identified is a rice plot based on the prediction result.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the rice plot identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the steps of the rice plot identification method according to any one of claims 1 to 7 are implemented.

Images