A method for acquiring soybean identification index image data, training and using a soybean spatial distribution mapping model, and related products.
By combining the soybean identification index model with multispectral features, the problem of separating soybeans from other crops in the same season was solved, achieving high-precision soybean remote sensing mapping with good stability and efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING NORMAL UNIVERSITY
- Filing Date
- 2025-07-09
- Publication Date
- 2026-06-30
AI Technical Summary
Existing technologies are insufficient to effectively improve the separability of soybeans from other crops growing at the same time, resulting in inadequate accuracy in soybean remote sensing mapping.
A novel soybean identification index, NSII, was developed by combining visible light, near-infrared, red-edge, and short-wave infrared features. The optimal segmentation threshold was determined through image segmentation and iterative optimization, thereby improving the separability of soybeans from other crops grown at the same time.
It improves the accuracy and efficiency of soybean remote sensing mapping, can be stably applied under different planting structures and climatic conditions, and has strong spatiotemporal mobility and mid-season mapping potential.
Smart Images
Figure CN120472331B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of crop identification technology, and in particular to a method for acquiring soybean identification index image data, training and using a soybean spatial distribution mapping model, and related products. Background Technology
[0002] Accurate spatial distribution of crops is crucial for monitoring and assessing farmland conditions, precision agriculture management, agricultural disaster response, and agricultural policy formulation. Traditional agricultural surveys for crop classification and mapping are typically time-consuming and labor-intensive, while remote sensing technology, with its macroscopic and rapid data acquisition capabilities, has become an important tool for crop mapping, significantly saving manpower, material resources, financial resources, and time costs, effectively compensating for the shortcomings of traditional survey methods. In recent years, the application of high-resolution (temporal and spatial) remote sensing data in crop identification has become increasingly widespread.
[0003] Spectral characteristics vary among crop types, providing crucial information for fine crop classification and thus becoming the most fundamental feature for crop type identification, playing a vital role in crop identification. Vegetation indices, derived through mathematical operations on multiple bands of remote sensing imagery, can enhance vegetation spectral characteristics and are frequently used in crop type identification and distribution mapping studies. Some researchers also focus on developing spectral indices specifically for identifying particular crop types, aiming to enhance the specific spectral characteristics of that crop while reducing the amount of remote sensing data input and simplifying the calculation process.
[0004] Accurate identification of soybean planting areas is crucial for ensuring food security. Dynamic classification mapping and monitoring of soybean spatial distribution provide vital data for agricultural decision-making. Farmers and agricultural policymakers can adjust planting plans based on this data, optimizing soybean planting layout and resource allocation to achieve optimal soybean production benefits. Therefore, many researchers are focusing on the development of soybean remote sensing identification indices.
[0005] However, soybeans and corn are planted almost simultaneously in many regions, and their phenological stages are highly similar, making them easily confused. Therefore, there is an urgent need in this field for a technical solution that can effectively improve the separability of soybeans from other concurrent crops. Summary of the Invention
[0006] The purpose of this application is to provide a method and related products for acquiring soybean identification index image data, training and using a soybean spatial distribution mapping model, and utilizing the soybean identification index image data to effectively improve the separability of soybeans from other crops growing at the same time.
[0007] To achieve the above objectives, this application provides the following solution:
[0008] Firstly, this application provides a method for acquiring soybean identification index image data, including:
[0009] Soybean identification index image data is obtained using a soybean identification index model; the soybean identification index model is as follows:
[0010] ;
[0011] ;
[0012] in, NSII This represents the calculated soybean identification index image data; EVI Indicates the enhanced vegetation index; Blue Re d , RE2 , NIR and SWIR 1 represents the satellite's blue, red, red-edge 2, near-infrared, and shortwave infrared 1 bands, respectively; G =2.5; C 1 =6; C 2 =7.5; L =1; The soybean identification index model is used for mapping the spatial distribution of soybeans.
[0013] Optionally, Blue Re d , RE2 , NIR and SWIR The specific bands of 1 are: B2, B4, B6, B8 and B11.
[0014] Secondly, this application provides a training method for a soybean spatial distribution mapping model, including:
[0015] Acquire training data; the training data includes: satellite multispectral imagery and sample data; the sample data is sample data that has been distinguished as soybeans and non-soybeans; the sample data is soybean and non-soybean samples in the monitoring area during the preset monitoring period, obtained through open and shared sample databases, existing high-precision soybean mapping products, field surveys, and digital interpretation of high-resolution remote sensing images.
[0016] Using the satellite multispectral imagery as input, the soybean identification index imagery data is calculated using the soybean identification index model; the soybean identification index model is the soybean identification index model described above.
[0017] A segmentation threshold is determined, and the soybean identification index image data is segmented using the segmentation threshold to obtain the image segmentation result;
[0018] Based on the soybean spatial distribution map, the F1 score of the image segmentation result is calculated, and the process of "determining the segmentation threshold and using the segmentation threshold to perform image segmentation on the soybean identification index image data to obtain the image segmentation result" is iterated until the F1 score is the same in two adjacent iterations, thus obtaining the optimal segmentation threshold; the optimal segmentation threshold is the optimal segmentation threshold of the soybean spatial distribution mapping model.
[0019] Optionally, when determining the segmentation threshold, the initial segmentation threshold is T1, where T1 = (T min +T max ) / 2; where T min T represents the minimum value of cultivated land pixels in soybean recognition index image data. max This represents the maximum value of cultivated land pixels in the soybean identification index image data.
[0020] Optionally, in the iterative process of calculating the F1 score of the image segmentation result based on the soybean spatial distribution map and returning the result, the method for determining the segmentation threshold for the next iteration step includes:
[0021] Based on the sample data, calculate the precision and recall of the image segmentation result in the current iteration step;
[0022] When precision is greater than recall, the segmentation threshold for the next iteration is T3; T3 = (T2 + T max T2 / 2, where T2 is the segmentation threshold in the current iteration step;
[0023] When precision is less than recall, the segmentation threshold for the next iteration is T4; T4 = (T min +T2) / 2.
[0024] Optionally, the formula for calculating the F1 score is:
[0025] ;
[0026] in, F1 Indicates the F1 score; P Indicates precision, R This indicates the recall rate.
[0027] Thirdly, this application provides a method for using a soybean spatial distribution mapping model, including:
[0028] Acquire satellite multispectral imagery of the area to be monitored;
[0029] Using the satellite multispectral imagery as input, the soybean identification index imagery data of the area to be monitored is calculated using the soybean identification index model; the soybean identification index model is the soybean identification index model described above.
[0030] Using the optimal segmentation threshold of the soybean spatial distribution mapping model, the soybean identification index image data of the area to be monitored is segmented to obtain a soybean spatial distribution map; the soybean spatial distribution mapping model is trained by the training method of the soybean spatial distribution mapping model described above.
[0031] Fourthly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the soybean identification index image data acquisition method or the soybean spatial distribution mapping model training method or the soybean spatial distribution mapping model usage method described above.
[0032] Fifthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the soybean identification index image data acquisition method or the soybean spatial distribution mapping model training method or the soybean spatial distribution mapping model usage method described above.
[0033] Sixthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the soybean identification index image data acquisition method or the soybean spatial distribution mapping model training method or the soybean spatial distribution mapping model usage method described above.
[0034] According to the specific embodiments provided in this application, the following technical effects are disclosed:
[0035] This application provides a method and related products for acquiring soybean identification index image data, training and using a soybean spatial distribution mapping model, and the image data acquisition method includes: acquiring soybean identification index image data using a soybean identification index model; the soybean identification index model is: ; ;in, NSII This represents the calculated soybean identification index image data; EVI Indicates the enhanced vegetation index; Blue Re d , RE2 , NIR and SWIR 1 represents the satellite's blue, red, red-edge 2, near-infrared, and shortwave infrared 1 bands, respectively; G =2.5; C 1 =6; C2 =7.5; L =1; The soybean identification index model is used for soybean spatial distribution mapping. This application proposes a novel soybean identification index model by comprehensively utilizing the features of visible light, near-infrared, red-edge, and short-wave infrared. This index model can improve the separability of soybeans from other crops in the same period, providing a new identification feature for soybean remote sensing mapping and helping to improve the accuracy of soybean distribution mapping. Attached Figure Description
[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0037] Figure 1 This diagram illustrates the application environment of the soybean identification index image data acquisition method, soybean spatial distribution mapping model training method, or usage method provided in the embodiments of this application.
[0038] Figure 2 This is a flowchart illustrating a training method for a soybean spatial distribution mapping model provided in an embodiment of this application.
[0039] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0040] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0041] This application utilizes features from visible light, near-infrared, red-edge, and short-wave infrared to develop a novel soybean identification index. This index improves the separability of soybeans from other crops growing at the same time, providing a new identification feature for soybean remote sensing mapping and helping to improve the accuracy of soybean distribution mapping.
[0042] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0043] The soybean identification index image data acquisition method, soybean spatial distribution mapping model training method, or usage method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on another server.
[0044] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, and tablets. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be a cloud server.
[0045] Example 1:
[0046] In an exemplary embodiment, a method for acquiring soybean identification index image data is provided. This method is executed by a computer device, specifically a terminal or server, or both. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps:
[0047] Soybean identification index image data is obtained using a soybean identification index model; the soybean identification index model is as follows:
[0048] ;
[0049] ;
[0050] in, NSII This represents the calculated soybean identification index image data; EVI Indicates the enhanced vegetation index; Blue Re d , RE2 , NIR and SWIR 1 represents the blue, red, red-edge 2, near-infrared, and shortwave infrared 1 bands of the Sentinel-2 satellite. For the Sentinel-2 satellite, the specific bands in this embodiment are the 2nd, 4th, 6th, 8th, and 11th bands, namely bands B2, B4, B6, B8, and B11. G =2.5; C 1 =6; C 2 =7.5; L =1; The soybean identification index model is used for mapping the spatial distribution of soybeans.
[0051] In addition, this embodiment also provides a method for constructing a soybean identification index model, the specific steps of which are as follows:
[0052] 1. Sample Data Acquisition: Based on the farmland data layer (CDL) of the most recent year in the monitoring area downloaded from the official US website, the areas with an annual cloud cover of less than 15% in the Sentinel-2 multispectral image were masked to the areas of the CDL crop layer with a confidence level greater than 95%. Then, random points were selected using the masked CDL data. These random points represent different crop types (soybeans, corn, wheat, cotton, and rice). The random points of different crop types were merged into two types: soybeans and non-soybeans. These random points were used as sample data.
[0053] 2. Image Acquisition: Sentinel-2 multispectral images of the monitoring area within the corresponding year of CDL were acquired using the Google Earth Engine platform. Cloud removal was performed using the QA band, and the images were then resampled to 10-meter resolution. Median composite images were then generated at 10-day intervals.
[0054] 3. Feature Index Calculation: Based on Sentinel-2 multispectral composite imagery, 35 remote sensing indices were calculated. The specific formulas are shown below, where... Bi This indicates the Sentinel-2's... i Each band.
[0055] Normalized Difference Vegetation Index (NDVI): .
[0056] Normalized Difference Red-edge 1 Index (NDre1): .
[0057] Normalized Difference Red-edge 2 Index (NDre2): .
[0058] Surface Moisture Index (LSWI): .
[0059] Soil-modified vegetation index (SAVI): .
[0060] Normalized Differential Moisture Index (NDWI): .
[0061] Plant Senescence Reflection Index (PSRI): .
[0062] Normalized Difference Vegetation Index (NDVIre1) Red Edge 1: .
[0063] Normalized Difference Vegetation Index Red Edge 2 (NDVIre2): .
[0064] Novel Inverted Red-Edge Chlorophyll Index (IRECI): .
[0065] Cire index (red-edged chlorophyll index): .
[0066] Ratio Vegetation Index (RVI): .
[0067] Wide Dynamic Range Vegetation Index (WDRVI): .
[0068] Nonlinear Vegetation Index (NLI): .
[0069] Modified Nonlinear Vegetation Index (MNLI): .
[0070] Optimize the soil-regulated vegetation index (OSAVI): .
[0071] Enhanced Vegetation Index (EVI): .
[0072] Differential Vegetation Index (DVI): .
[0073] Bare Soil Index (BSI): .
[0074] Winter rapeseed index (WRI): .
[0075] Green light ratio vegetation index (RVI) Green ): .
[0076] The ratio of red light to red edge bands in the vegetation index (SR) Red / Green ): .
[0077] Visible light atmospheric impedance index (VARI) Green ): .
[0078] Triangular Vegetation Index (TVI): , .
[0079] MERIS Terrestrial Chlorophyll Index (MTCI): .
[0080] Modified Normalized Difference Vegetation Index (MNDVI): .
[0081] Modified Simple Ratio Vegetation Index (MSAVI): .
[0082] Corrected Simple Ratio Red Edge Index (MSRre): .
[0083] Corrected Simple Ratio Narrow Red Edge Index (MSRren): .
[0084] Renormalized Difference Vegetation Index (RDVI): .
[0085] Simple ratio (SR): .
[0086] Red border position (REP): .
[0087] Normalized Burning Index (NBR): .
[0088] Greenness and Moisture Composite Index (GWCCI): .
[0089] Soybean Graphical Composite Index (SMCI): , , ,in, GVCI This represents the green chlorophyll vegetation index.
[0090] 4. Feature Separability Analysis and Determination of Optimal Monitoring Time Window: The JM (Jffries-Matusita) distance was used as the separation index to analyze the separability of soybeans from other crops in the same period on different features. The JM distance ranges from 0 to 2; the greater the difference in features between the two crops, the larger the JM distance. Different features include 10 bands of Sentinel-2, specifically bands 2, 3, 4, 5, 6, 7, 8, 8A, 11, and 12, i.e., bands B2, B3, B4, B5, B6, B7, B8, B8A, B11, and B12, as well as the 35 remote sensing indices calculated in the previous step. Comparing the JM distances calculated based on different features at different times, a larger JM distance indicates higher separability. The feature and period with the highest separation were identified, and the period with the highest separation was determined as the optimal monitoring window.
[0091] 5. Determination of the Soybean Identification Index Model NSII: Adhering to the principle of fully utilizing and minimizing feature redundancy, within the optimal monitoring window, arithmetic operations such as addition, subtraction, multiplication, division, and reciprocal transformations are used to combine features with high separation from 10 spectral bands and 35 indices to form a new index. The 10 spectral bands with high separation include the red-edge bands (RE2, RE3, RE4), the near-infrared band (NIR), and the short-wave infrared bands (SWIR1 and SWIR2). Among the three red-edge bands, RE2 performs best; among the two short-wave infrared bands, SWIR1 performs better. The 35 indices with high separation include SMCI, GWCCI, and EVI. SMCI uses 8 bands and already considers EVI; GWCCI is the product of SWIR1 and NDVI, utilizing red, near-infrared, and short-wave infrared bands. To comprehensively utilize information from different spectral bands while reducing feature redundancy, a novel soybean identification index (NSII) was constructed by selecting RE2 from three red-edge bands, SWIR1 from two shortwave infrared bands, and the EVI index from three indices (SMCI, GWCCI, and EVI) which provides supplementary information in the red and near-infrared bands but does not involve red-edge or shortwave infrared in the calculation. Mathematical processing (such as addition, subtraction, multiplication, division, and reciprocal transformations) was applied to the three basic features RE2, SWIR1, and EVI to enhance the soybean's characteristic information. Finally, the arithmetic operation that yielded the highest average separation within the optimal time window was selected as the final NSII calculation formula. Details are as follows:
[0092] ;
[0093] ;
[0094] in, NSII This represents the calculated soybean identification index image data; EVI Indicates the enhanced vegetation index; Blue Re d , RE2 , NIR and SWIR 1 refers to the blue, red, red-edge 2, near-infrared and shortwave infrared 1 bands of the Sentinel-2 satellite, specifically the 2nd, 4th, 6th, 8th and 11th bands, namely bands B2, B4, B6, B8 and B11; G =2.5; C 1 =6; C 2 =7.5; L =1.
[0095] Compared with related technologies, the soybean identification index model (NSII index) in this embodiment has the following advantages:
[0096] 1) It features lightweight design. It employs a simple and easy-to-understand calculation process. By filtering spectral bands and remote sensing indices, it efficiently combines bands and indices with high JM distances in soybean identification, reducing the complexity of the calculation process, saving data processing time and computing resources, and thus improving the overall efficiency of mapping.
[0097] 2) It has better soybean identification capabilities. Analysis of the JM distance shows that the separation of NSII > the separation of the Soybean Mapping Composite Index (SMCI) > the separation of the Greenness and Moisture Composite Index (GWCCI). Furthermore, the soybean identification accuracy based on NSII is higher than that based on SMCI.
[0098] 3) It has good stability and portability. The NSII index has been tested for three years in multiple regions with different planting structures and climatic conditions, and has achieved good accuracy (all above 80%), indicating that it has strong spatiotemporal mobility.
[0099] 4) It has the potential for mid-season mapping. NSII does not require data for the entire growing season; as long as data is obtained within the optimal time window, soybean distribution mapping can begin. Moreover, the optimal time window is relatively long, which is conducive to the collection of high-quality data and improves the practical application capability of NSII to a certain extent.
[0100] Example 2:
[0101] like Figure 2 As shown in the embodiments of this application, a training method for a soybean spatial distribution mapping model is also provided, including the following steps:
[0102] S1. Acquire training data; the training data includes: satellite multispectral imagery and sample data; the sample data consists of samples that have been distinguished as soybeans and non-soybeans; the sample data consists of soybean and non-soybean samples from the monitoring area during the preset monitoring period, obtained through open and shared sample databases, existing high-precision soybean mapping products, field surveys, and high-resolution remote sensing image visual interpretation digitization. In this embodiment, the sample data consists of soybean and non-soybean samples from the monitoring area during the preset monitoring period, obtained through open and shared sample databases, existing high-precision soybean mapping products, field surveys, and high-resolution remote sensing image visual interpretation digitization. The sample data in this embodiment, obtained through existing high-precision soybean mapping products, consists of soybean and non-soybean samples from the monitoring area during the preset monitoring period, used to calculate the F1 score of the image segmentation results for training.
[0103] The obtained samples were randomly divided into two parts in a 7:3 ratio, with 70% used for threshold determination and 30% used for accuracy verification.
[0104] Sentinel-2 multispectral images of soybeans within the optimal monitoring time window of the monitoring area in the monitoring year were obtained using the Google Earth Engine platform. Cloud removal was performed using the QA band, and the images were then resampled to a 10-meter resolution. Subsequently, the median values of the images were composited to generate a 10-meter resolution median composite image within the optimal monitoring time window.
[0105] S2. Using the satellite multispectral imagery of Sentinel 2 as input, the soybean identification index image data is calculated using the soybean identification index model; the soybean identification index model is the soybean identification index model described in Example 1.
[0106] Using the median composite image with a resolution of 10 meters obtained in the previous step as input, the NSII within the optimal monitoring time window for soybeans is calculated according to the NSII calculation formula (soybean identification index model) to obtain the NSII image (soybean identification index image data).
[0107] S3. Determine the segmentation threshold and use the segmentation threshold to perform image segmentation on the soybean identification index image data to obtain the image segmentation result.
[0108] S4. Based on the sample data, calculate the F1 score of the image segmentation result, and return to the step of "determine the segmentation threshold, and use the segmentation threshold to perform image segmentation on the soybean identification index image data to obtain the image segmentation result" for iteration until the F1 score in two adjacent iterations is the same, and obtain the optimal segmentation threshold; the optimal segmentation threshold is the optimal segmentation threshold of the soybean spatial distribution mapping model.
[0109] When determining the segmentation threshold, the initial segmentation threshold is T1, where T1 = (T min +T max ) / 2; where T min T represents the minimum value of cultivated land pixels in soybean recognition index image data. max This represents the maximum value of cultivated land pixels in the soybean identification index (NSII) image data. Cultivated land pixels refer to the pixels representing cultivated land in the image, excluding buildings and other structures. This segmentation threshold is used to divide the NSII image into two parts.
[0110] Methods for determining the segmentation threshold for the next iteration step during the iteration process include:
[0111] Based on the soybean spatial distribution map, calculate the precision and recall of the image segmentation results in the current iteration step.
[0112] When precision is greater than recall (indicating that misclassifications outweigh missedclassifications, resulting in a lower threshold), the segmentation threshold for the next iteration is T3; T3 = (T2 + T max ) / 2, where T2 is the segmentation threshold in the current iteration step.
[0113] When precision is less than recall (indicating that missed classifications outnumber misclassifications, resulting in a higher threshold), the segmentation threshold for the next iteration is T4; T4 = (T min +T2) / 2.
[0114] Then, compare the F1 scores before and after updating the threshold. If the F1 score improves, continue iterating using the same logic. Repeat this process until the optimal segmentation threshold is obtained. Based on the optimal segmentation threshold, perform NSII image segmentation to obtain the final soybean distribution map.
[0115] The formula for calculating the F1 score is as follows:
[0116] ;
[0117] in, F1 Indicates the F1 score; P Indicates precision, R This indicates the recall rate.
[0118] Finally, this embodiment also provides an accuracy verification process:
[0119] Using the remaining 30% of the sample, calculate the error matrix to verify the accuracy of the soybean distribution map. The error matrix is calculated as follows:
[0120] In the error matrix table, n represents the number of categories. The columns of the error matrix represent the reference image, and the rows represent the number of samples where there is misclassification or other issues between the evaluated image and the corresponding category of the reference image.
[0121] The main diagonal element (x) in the table 11 ,x 22 ,…,x nn The number of correctly classified samples is represented by ), while the number of samples misclassified relative to the reference image is represented by the number of samples misclassified. The row total represents the total number of samples of that category in the evaluated image, and the column total represents the total number of samples of that category in the reference image, as shown in the table below.
[0122] Table 1 Error Matrix
[0123]
[0124] Based on the error matrix, a series of accuracy evaluation indicators are calculated to evaluate the classification and extraction results. The basic evaluation indicators are as follows:
[0125] (1) Overall classification accuracy:
[0126] ;
[0127] Wherein, OA represents the overall classification accuracy. x kk Let represent the number of correctly classified samples in class k, n represent the number of correctly classified sample classes, and N represent the total number of samples. Overall classification accuracy is the ratio of the number of correctly classified samples to the total number of samples, and it is an important indicator in accuracy assessment. The number of correctly classified samples is the sum of the elements on the main diagonal of the error matrix, while the total number of samples represents the sum of all samples.
[0128] (2) User precision (for class i):
[0129] ;
[0130] Where UA represents user precision. x ii Indicates the first i The number of samples in the class that match the category of the reference image. x i+ Indicates the first data in the evaluated data i The total number of samples in a class; user precision represents the ratio of the number of samples correctly classified into that class to the total number of samples in that class in the evaluated data.
[0131] (3) Cartographic accuracy (for class i):
[0132] ;
[0133] Here, PA represents the mapping accuracy. x jj This indicates that the entire image was correctly classified to the [number]th [level]. j Number of samples in the class x +j Indicates the first in the reference image j The total number of samples of a class; mapping accuracy refers to the ratio of the number of samples correctly classified into that class in the entire image to the total number of samples of that class in the reference image.
[0134] (4) Kappa coefficient:
[0135] ;
[0136] In the formula, n represents the total number of land cover categories in the evaluated data; x ii The values on the main diagonal of the error matrix; x i+ andx +i , respectively, represent the total number of samples in the i-th row and i-th column; N is the total number of samples in the accuracy assessment. The Kappa coefficient ranges from 0 to 1, with a value closer to 1 indicating better consistency.
[0137] Example 3:
[0138] This application also provides a method for using a soybean spatial distribution mapping model, including the following steps:
[0139] Acquire satellite multispectral images of the area to be monitored; in this embodiment, Sentinel-2 satellite multispectral images are acquired.
[0140] Using the satellite multispectral imagery as input, the soybean identification index imagery data of the area to be monitored is calculated using the soybean identification index model; the soybean identification index model is the soybean identification index model described above.
[0141] Using the optimal segmentation threshold of the soybean spatial distribution mapping model, the soybean identification index image data of the area to be monitored is segmented to obtain a soybean spatial distribution map; the soybean spatial distribution mapping model is trained by the training method of the soybean spatial distribution mapping model described in Example 2.
[0142] It should be noted that in this embodiment, when the satellite multispectral imagery changes, the soybean spatial distribution mapping model needs to be retrained. The changes in the satellite multispectral imagery mainly involve changes in the location of the monitoring area and changes in time.
[0143] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for acquiring soybean identification index image data, a training method for a soybean spatial distribution mapping model, or a method for using the soybean spatial distribution mapping model.
[0144] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0145] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.
[0146] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0147] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0149] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
[0150] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0151] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0152] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A training method for a soybean spatial distribution mapping model, characterized in that, include: Acquire training data; the training data includes: satellite multispectral imagery and sample data; the sample data is sample data that has been distinguished as soybeans and non-soybeans; the sample data is soybean and non-soybean samples in the monitoring area during the preset monitoring period, obtained through open and shared sample databases, existing high-precision soybean mapping products, field surveys, and digital interpretation of high-resolution remote sensing images. Using the satellite multispectral imagery as input, the soybean identification index imagery data is calculated using the soybean identification index model; A segmentation threshold is determined, and the soybean identification index image data is segmented using the segmentation threshold to obtain the image segmentation result; Based on the sample data, the F1 score of the image segmentation result is calculated, and the process of "determining the segmentation threshold and using the segmentation threshold to perform image segmentation on the soybean identification index image data to obtain the image segmentation result" is iterated until the F1 score is the same in two adjacent iterations, thus obtaining the optimal segmentation threshold; the optimal segmentation threshold is the optimal segmentation threshold of the soybean spatial distribution mapping model. The soybean identification index model is as follows: ; ； in, NSII This represents the calculated soybean identification index image data; EVI Indicates the enhanced vegetation index; Blue Re d , RE2 , NIR and SWIR 1 represents the satellite's blue, red, red-edge 2, near-infrared, and shortwave infrared 1 bands, respectively; G =2.5; C 1 =6; C 2 =7.5; L =1; The soybean identification index model is used for mapping the spatial distribution of soybeans; Blue Re d , RE2 , NIR and SWIR The specific bands of 1 are the 2nd, 4th, 6th, 8th and 11th bands of Sentinel 2, namely bands B2, B4, B6, B8 and B11; When determining the segmentation threshold, the initial segmentation threshold is T1, where T1 = (T min +T max ) / 2; where T min T represents the minimum value of cultivated land pixels in soybean recognition index image data. max This represents the maximum value of cultivated land pixels in the soybean identification index image data; Methods for determining the segmentation threshold for the next iteration step during the iteration process include: Based on the sample data, calculate the precision and recall of the image segmentation result in the current iteration step; When precision is greater than recall, the segmentation threshold for the next iteration is T3; T3 = (T2 + T max T2 / 2, where T2 is the segmentation threshold in the current iteration step; When precision is less than recall, the segmentation threshold for the next iteration is T4; T4 = (T min +T2) / 2.
2. The training method for the soybean spatial distribution mapping model according to claim 1, characterized in that, The formula for calculating the F1 score is: ; in, F1 Indicates the F1 score; P Indicates precision, R This indicates the recall rate.
3. A method for using a soybean spatial distribution mapping model, characterized in that, include: Acquire satellite multispectral imagery of the area to be monitored; Using the satellite multispectral imagery as input, the soybean identification index image data of the area to be monitored is calculated using the soybean identification index model; Using the optimal segmentation threshold of the soybean spatial distribution mapping model, the soybean identification index image data of the area to be monitored is segmented to obtain a soybean spatial distribution map; the soybean spatial distribution mapping model is trained by the training method of the soybean spatial distribution mapping model according to claim 1 or 2.
4. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the training method of the soybean spatial distribution mapping model according to claim 1 or 2, or the method of using the soybean spatial distribution mapping model according to claim 3.
5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the training method of the soybean spatial distribution mapping model according to claim 1 or 2, or the method of using the soybean spatial distribution mapping model according to claim 3.
6. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the training method of the soybean spatial distribution mapping model according to claim 1 or 2, or the method of using the soybean spatial distribution mapping model according to claim 3.