A soil type map updating method and system
Through the combination of support vector machine and geo-weighted regression algorithm, the problems of low classification accuracy and blurred boundaries in soil type map update are solved, and high-precision classification of soil types are realized, suitable for multi-spectral remote sensing image data, and the soil type is improved The efficiency and automation of the graph update.
Patent Information
- Application Number
- CN202411125128.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2044-08-16
AI Technical Summary
The existing soil type map update methods have problems such as low classification accuracy, blurred boundaries and high cost of obtaining labeled data, making it difficult to effectively use multispectral remote sensing image data for accurate updates.
The support vector machine classification model is used to initially classify multispectral remote sensing image data, and combine it with the geographical weighted regression algorithm for refined classification, making full use of the spectral and spatial information of the image data to improve the accuracy of soil type classification.
By combining support vector machine and geo-weighted regression algorithm, preliminary classification and refined classification of soil types are achieved, classification accuracy is improved, boundaries are clearer, suitable for multiple types of multispectral remote sensing image data, and soil type map update is improved efficiency and automation.
Smart Images
Figure CN118658067B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of image processing, and in particular relates to a soil type map updating method and system. Background Art
[0002] Soil type maps are one of the important basic data in the fields of agriculture, environmental protection and land use planning. In the fields of geography and soil science, the compilation and updating of soil type maps are of great significance for agricultural planning, environmental management and land use. Accurate soil type maps can provide scientific guidance for agricultural production, help farmers choose appropriate crops and fertilization plans, and improve agricultural production efficiency; in terms of environmental protection, soil type maps can be used to monitor and evaluate soil quality, prevent and control soil pollution; in land use planning, soil type maps can provide a basis for the rational development and utilization of land resources.
[0003] Traditional soil type map production methods mainly rely on field sampling and laboratory analysis. Although this method has high accuracy, it has problems such as long time consumption, high cost, and limited coverage. With the rapid development of remote sensing technology, multispectral remote sensing image data has gradually become an important data source for soil type identification and classification due to its rich spectral information and high spatial resolution. However, the complexity and diversity of remote sensing image data make the automatic classification and identification of soil types a challenging task. Relying solely on multispectral remote sensing image data for soil type classification often has problems such as low classification accuracy and blurred boundaries. Existing soil type map update methods mainly include classification methods based on supervised learning and clustering methods based on unsupervised learning. Supervised learning methods require a large amount of labeled data for training. Although the classification accuracy is high, the cost of obtaining labeled data is high. Unsupervised learning methods do not require labeled data, but the classification accuracy is low and it is difficult to explain the classification results.
[0004] Therefore, how to effectively utilize multispectral remote sensing image data and combine it with advanced classification algorithms to achieve accurate updating of soil type maps is a technical problem that needs to be solved urgently. Summary of the invention
[0005] The purpose of the present invention is to provide a soil type map updating method and system, which can make full use of the spectral information and spatial information of the image data and improve the accuracy of soil type classification by using a support vector machine classification model to perform preliminary classification of multispectral remote sensing image data and then combining a geographically weighted regression algorithm to perform fine classification of the preliminary classification results.
[0006] The specific technical solutions are as follows:
[0007] In a first aspect, the present invention provides a soil type map updating method, the method comprising the following steps:
[0008] Step S1, acquiring multispectral remote sensing image data of a target area, and preprocessing the multispectral remote sensing image data to obtain preprocessed image correction data.
[0009] Step S2, based on the image correction data, using a support vector machine classification model to preliminarily classify the soil types to obtain a preliminary classification result.
[0010] Step S3, based on the preliminary classification results, fine classification of soil types is performed according to a geographically weighted regression algorithm to obtain a fine classification result.
[0011] Step S4, updating the soil type map according to the refined classification results.
[0012] Furthermore, the multispectral remote sensing image data includes: data of red light band, near infrared band and short-wave infrared band; the preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal.
[0013] Furthermore, the support vector machine classification model is used to perform preliminary classification on the preprocessed image data, including:
[0014] Select historical soil sample information to construct a training sample set, which is represented as: ;in is the feature vector of the image data, is the corresponding soil type label, is the number of training samples; the feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the shortwave infrared band.
[0015] The kernel function of the classification model It is expressed as:
[0016] ;
[0017] The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as:
[0018] ;
[0019] in, are kernel function parameters, is the Lagrange multiplier, is a regularization parameter used to control the complexity and error of the model.
[0020] The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results:
[0021] ;
[0022] is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term on the decision boundary.
[0023] Furthermore, the kernel function parameters are selected by cross-validation. and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as training sets; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets.
[0024] Select the optimal kernel function parameter σ and regularization parameter C according to the F1 value performance index of each verification;
[0025] ;
[0026] in, Recall represents the precision and Recall represents the recall.
[0027] Furthermore, based on the preliminary classification results, the geographically weighted regression algorithm was used to refine the soil type classification, specifically:
[0028] Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight:
[0029] ;
[0030] in, It's a pixel and The distance between is the bandwidth parameter.
[0031] Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients :
[0032] ;
[0033] in, is the feature matrix, This is the preliminary classification result.
[0034] Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results .
[0035] In a second aspect, the present invention provides a soil type map updating system, the system comprising: a data acquisition and processing module, a preliminary classification module, a fine classification module and a type map updating module, each module being connected in sequence.
[0036] The data acquisition and processing module is used to acquire multispectral remote sensing image data of a target area, and preprocess the multispectral remote sensing image data to obtain preprocessed image correction data.
[0037] The preliminary classification module is used to perform preliminary classification of soil types based on the image correction data using a support vector machine classification model to obtain preliminary classification results.
[0038] The fine classification module is used to perform fine classification of soil types based on the preliminary classification results according to the geographically weighted regression algorithm to obtain fine classification results.
[0039] The type map updating module is used to update the soil type map according to the refined classification results.
[0040] Furthermore, the multispectral remote sensing image data acquired by the data acquisition and processing module includes: data of red light band, near infrared band and short-wave infrared band; the preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal.
[0041] Furthermore, the preliminary classification module includes a support vector machine classification model, and the support vector machine classification model is used to perform preliminary classification on the preprocessed image data, specifically including:
[0042] The historical soil sample information is selected to construct the training sample set, which is represented as ;in is the feature vector of the image data, is the corresponding soil type label, is the number of training samples; the feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the shortwave infrared band.
[0043] The kernel function of the classification model It is expressed as:
[0044] ;
[0045] The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as:
[0046] ;
[0047] in, are kernel function parameters, is the Lagrange multiplier, is a regularization parameter used to control the complexity and error of the model.
[0048] The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results:
[0049] ;
[0050] is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term on the decision boundary.
[0051] Furthermore, the kernel function parameters are selected by cross-validation. and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as training sets; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets.
[0052] According to the F1 value performance index of each verification; select the optimal kernel function parameter σ and regularization parameter C;
[0053] ;
[0054] in, Recall represents the precision and Recall represents the recall.
[0055] Furthermore, based on the preliminary classification results, the geographically weighted regression algorithm was used to refine the soil type classification, specifically:
[0056] Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight:
[0057] ;
[0058] in, It's a pixel and The distance between is the bandwidth parameter.
[0059] Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients :
[0060] ;
[0061] in, is the feature matrix, This is the preliminary classification result.
[0062] Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results .
[0063] Compared with the prior art, the present invention has the following beneficial effects:
[0064] The present invention realizes preliminary classification and refined classification of soil types by combining a support vector machine classification model and a geographically weighted regression algorithm, effectively improving the classification accuracy. The geographically weighted regression algorithm takes into account the influence of spatial position on the classification result, making the classification boundary clearer and more accurate. The entire method flow of the present invention includes data preprocessing, preliminary classification, refined classification and soil type map updating, has a high degree of automation, can quickly process a wide range of remote sensing image data, and improves the efficiency of soil type map updating. The present invention is applicable to various types of multispectral remote sensing image data, including data in the red light band, near infrared band and short-wave infrared band, and has strong applicability. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] Figure 1 The present invention is a flow chart of a soil type map updating method.
[0066] Figure 2 The present invention is a schematic diagram of the composition of a soil type map updating system. DETAILED DESCRIPTION
[0067] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are 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.
[0068] Example 1: Figure 1 FIG. 1 is a flow chart of a soil type map updating method of the present invention, wherein the method comprises the following steps:
[0069] Step S1, acquiring multispectral remote sensing image data of a target area, and preprocessing the multispectral remote sensing image data to obtain preprocessed image correction data.
[0070] It should be noted that multispectral remote sensing image data include: data from the red light band, near infrared band and short-wave infrared band; multispectral remote sensing image data are usually obtained by satellites, drones or other remote sensing platforms. These data contain the reflection or radiation information of the target area in different bands. For soil type classification, the data of the red light band, near infrared band and short-wave infrared band are particularly important. The red light band usually corresponds to the chlorophyll absorption characteristics of vegetation, which is helpful in distinguishing vegetation cover types and indirectly affects the identification of soil types; the near infrared band is an important identification band for vegetation and water bodies, which is sensitive to soil moisture and vegetation cover density, and helps to distinguish different types of soil; the short-wave infrared band is sensitive to soil mineral composition, water content and organic matter content, and is an indispensable source of information in soil type classification.
[0071] Preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal. The purpose of radiation correction is to correct the radiation distortion caused by atmospheric conditions, sensor characteristics and solar altitude angle, so as to ensure that the image data can truly reflect the surface reflectivity or radiance; the purpose of atmospheric correction is to remove the influence of atmospheric absorption and scattering on the image data, especially to remove the interference of water vapor and aerosol in the atmosphere on the spectral characteristics, so as to improve the accuracy and reliability of the data; the purpose of geometric correction is to correct the geometric distortion of the image caused by factors such as sensor attitude, earth curvature and terrain undulation, so as to ensure the accuracy of the geographic coordinates in the image, and provide a basis for subsequent spatial analysis and classification; the purpose of noise removal is to remove random noise and outliers in the image through filtering, denoising and other algorithms, so as to improve the signal-to-noise ratio and clarity of the image, and provide a clean data source for subsequent classification processing.
[0072] Step S2: Based on the image correction data, a support vector machine classification model is used to perform a preliminary classification of soil types to obtain a preliminary classification result.
[0073] The support vector machine classification model is used to perform preliminary classification on the preprocessed image data, including:
[0074] Select historical soil sample information to construct a training sample set, which is represented as: ;in is the feature vector of the image data, is a corresponding soil type label, and the soil type label includes at least: sandy soil, clay soil, and loam; is the number of training samples; the collected historical soil sample information, these samples should contain soil type labels and corresponding multispectral remote sensing image data, these samples can be soil samples collected in the field, and have been professionally classified and identified.
[0075] The feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the shortwave infrared band.
[0076] The support vector machine classification model measures the similarity between samples through kernel functions. Commonly used kernel functions include linear kernels, polynomial kernels, radial basis function (RBF) kernels, etc. Here, the radial basis function is used. The kernel function of the classification model is It is expressed as:
[0077] ;
[0078] The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as:
[0079] ;
[0080] in, are kernel function parameters, is the Lagrange multiplier, is a regularization parameter used to control the complexity and error of the model.
[0081] The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results:
[0082] ;
[0083] is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term on the decision boundary.
[0084] Furthermore, the kernel function parameters are selected by cross-validation. and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as training sets; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets.
[0085] Select the optimal kernel function parameter σ and regularization parameter C according to the F1 value performance index of each verification;
[0086] ;
[0087] in, Recall represents the precision and Recall represents the recall.
[0088] The calculation formula of precision is:
[0089] ;
[0090] in, (True Positive) is a true example, indicating the number of samples correctly predicted as positive; (False Positive): False positive examples, which means the number of samples that are incorrectly predicted as positive.
[0091] The calculation formula of recall rate Recall is:
[0092] ;
[0093] False Negative refers to the number of samples that are incorrectly predicted as negative.
[0094] Step S3, based on the preliminary classification results, fine classification of soil types is performed according to a geographically weighted regression algorithm to obtain a fine classification result.
[0095] Based on the preliminary classification results, the geographically weighted regression algorithm was used to refine the soil type classification, specifically:
[0096] Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight:
[0097] ;
[0098] in, It's a pixel and The distance between is the bandwidth parameter.
[0099] Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients :
[0100] ;
[0101] in, is the feature matrix, This is the preliminary classification result.
[0102] Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results .
[0103] Step S4, updating the soil type map according to the refined classification results.
[0104] The main steps in updating soil type maps include: data integration and mapping, soil type map compilation, verification and evaluation, and results publication and application.
[0105] Integrate the refined classification results (i.e., the soil type label for each pixel) with the corresponding geographic location information (such as longitude and latitude coordinates) to form a complete soil type dataset; use geographic information system (GIS) software or related spatial data processing tools to map the soil type dataset to the geographic space of the target area, convert the soil type label into an attribute field in the GIS layer, and assign corresponding colors or symbols to distinguish different soil types.
[0106] If the target area already has an old soil type map or other related layers (such as topographic maps, vegetation maps, etc.), the old soil type map or other related layers can be overlaid with the refined classification results for analysis in order to have a more comprehensive understanding of the spatial distribution and relationships of soil types. The specific method is: check whether the coordinate systems of all layers (including refined classification results, old soil type maps, topographic maps, vegetation maps, etc.) are consistent; if not, use GIS software (such as ArcGIS, QGIS, etc.) to convert the coordinates to ensure that they can be overlaid in the same geographic space framework; convert all layers to formats supported by GIS software, such as Shapefile, GeoTIFF, GeoJSON, etc.; use GIS software: open the GIS software, load all prepared layers, and use the overlay function of the GIS software (such as the "Overlay Analysis" tool in ArcGIS) to overlay the refined classification result layer with the old soil type map, topographic map, vegetation map, etc. one by one or simultaneously.
[0107] Select a certain number of sample points for field verification to evaluate the accuracy and reliability of the refined classification results; evaluate the performance of the refined classification model based on the results of the field verification and other relevant data sources; the evaluation indicators may include statistical indicators such as classification accuracy, recall rate, F1 score, and qualitative indicators such as the degree of consistency with the actual soil type distribution; publish the updated soil type map in an appropriate format (such as PDF, picture, GIS file, etc.).
[0108] Example 2: Figure 2 As shown, it is a schematic diagram of the composition of a soil type map updating system of the present invention. The soil type map updating system is an integrated processing platform, which aims to achieve accurate updating of the soil type map of the target area through multi-source data fusion and advanced algorithm application. The system consists of four core parts: data acquisition and processing module, preliminary classification module, fine classification module and type map updating module. The modules work closely together to ensure smooth data flow and efficient processing.
[0109] The system comprises: a data collection and processing module, a preliminary classification module, a fine classification module and a type map updating module, and each module is connected in sequence.
[0110] The data acquisition and processing module is used to obtain multispectral remote sensing image data of the target area, preprocess the multispectral remote sensing image data, and obtain preprocessed image correction data. This module is mainly responsible for obtaining multispectral remote sensing image data of the target area from satellites, drones or other remote sensing platforms, and then preprocessing the data, including radiation correction, geometric correction, atmospheric correction, noise removal, band fusion and other steps to eliminate various interference factors, improve data quality, and finally output preprocessed image correction data to provide a reliable data basis for subsequent classification.
[0111] The preliminary classification module is used to perform preliminary classification of soil types based on the image correction data using a support vector machine classification model to obtain preliminary classification results.
[0112] The fine classification module is used to perform fine classification of soil types based on the preliminary classification results according to the geographically weighted regression algorithm to obtain fine classification results.
[0113] The type map updating module is used to update the soil type map according to the refined classification results.
[0114] The multispectral remote sensing image data acquired by the data acquisition and processing module include: data of red light band, near infrared band and short-wave infrared band; the preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal.
[0115] The preliminary classification module includes a support vector machine classification model, and uses the support vector machine classification model to perform preliminary classification on the preprocessed image data, specifically including:
[0116] Select historical soil sample information to construct a training sample set, which is represented as: ;in is the feature vector of the image data, is the corresponding soil type label, is the number of training samples; the feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the shortwave infrared band.
[0117] The kernel function of the classification model It is expressed as:
[0118] ;
[0119] The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as:
[0120] ;
[0121] in, are kernel function parameters, is the Lagrange multiplier, is a regularization parameter used to control the complexity and error of the model.
[0122] The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results:
[0123] ;
[0124] is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term on the decision boundary.
[0125] Select kernel function parameters through cross-validation and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as training sets; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets.
[0126] According to the F1 value performance index of each verification; select the optimal kernel function parameter σ and regularization parameter C;
[0127] ;
[0128] in, Recall represents the precision and Recall represents the recall.
[0129] Based on the preliminary classification results, the geographically weighted regression algorithm was used to refine the soil type classification, specifically:
[0130] Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight:
[0131] ;
[0132] in, It's a pixel and The distance between is the bandwidth parameter.
[0133] Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients :
[0134] ;
[0135] in, is the feature matrix, This is the preliminary classification result.
[0136] Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results .
[0137] It should be noted that Reflects the pixel In the surrounding local area, the feature matrix The features in the preliminary classification results The impact of Each element in (corresponding to the feature matrix Each column in ) represents the intensity and direction of the influence of the corresponding feature on the classification result; a positive value means that the feature has a positive influence on the classification result, that is, when the feature value increases, the classification result tends to increase (or shift to a certain category); a negative value indicates a negative influence; the absolute value indicates the intensity of the influence.
[0138] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A soil type map updating method, characterized in that: The method comprises the following steps: Step S1, acquiring multispectral remote sensing image data of a target area, and preprocessing the multispectral remote sensing image data to obtain preprocessed image correction data; The multispectral remote sensing image data includes: data of red light band, near infrared band and short-wave infrared band; the preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal; Step S2, based on the image correction data, using a support vector machine classification model to preliminarily classify the soil types to obtain a preliminary classification result; Select historical soil sample information to construct a training sample set, which is represented as: ;in is the feature vector of the image data, is the corresponding soil type label, is the number of training samples; the feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the short-wave infrared band; The kernel function of the classification model It is expressed as: ； The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as: ； in, are kernel function parameters, is the Lagrange multiplier, is the regularization parameter, which is used to control the complexity and error of the model; The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results: ； is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term of the decision boundary; Select kernel function parameters through cross-validation and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as the training set; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets; the optimal kernel function parameter σ and regularization parameter C are selected according to the F1 value performance index of each validation; ； in, Represents precision, Recall represents recall; Step S3, based on the preliminary classification results, finely classify the soil types according to the geographically weighted regression algorithm to obtain a fine classification result; Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight: ； in, It's a pixel and The distance between is the bandwidth parameter; Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients : ； in, is the feature matrix, is the preliminary classification result; Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results ; Step S4, updating the soil type map according to the refined classification result; The steps to update soil type maps include: data integration and mapping, soil type map compilation, verification and evaluation, and results publication and application; Integrate the refined classification results with the corresponding geographic location information to form a complete soil type dataset; use the geographic information system to map the soil type dataset to the geographic space of the target area, convert the soil type labels into attribute fields in the GIS layer, and assign corresponding colors or symbols to distinguish different soil types; If the target area already has an old soil type map, the old soil type map should be overlaid with the refined classification results for analysis. The specific method is: check whether the coordinate systems of all layers are consistent; if not, use GIS software to convert the coordinates; convert all layers to a format supported by GIS software; and overlay the refined classification result layers with the old soil type map one by one through the overlay function of GIS software.
2. A soil type map updating system, characterized in that: The system comprises: a data collection and processing module, a preliminary classification module, a fine classification module and a type map updating module; The data acquisition and processing module is used to obtain multispectral remote sensing image data of the target area, preprocess the multispectral remote sensing image data, and obtain preprocessed image correction data; The multispectral remote sensing image data acquired by the data acquisition and processing module includes: data of red light band, near infrared band and short-wave infrared band; the preprocessing includes radiation correction, atmospheric correction, geometric correction and noise removal; The preliminary classification module is used to perform preliminary classification of soil types based on the image correction data using a support vector machine classification model to obtain preliminary classification results; Select historical soil sample information to construct a training sample set, which is represented as: ;in is the feature vector of the image data, is the corresponding soil type label, is the number of training samples; the feature vector Contains multispectral information of soil images: , Indicates The value of each sample in the red light band; Indicates The value of each sample in the near-infrared band; Indicates The value of each sample in the short-wave infrared band; The kernel function of the classification model It is expressed as: ； The support vector machine classification model is trained by solving an optimization problem, and the optimization problem and the constraints of the optimization problem are expressed as: ； in, are kernel function parameters, is the Lagrange multiplier, is the regularization parameter, which is used to control the complexity and error of the model; The preprocessed image data is classified using the trained support vector machine classification model to obtain the preliminary classification results: ； is the decision function of the SVM classifier, and the sign function is a symbolic function used to convert the output of the decision function into a category label. is the bias term of the decision boundary; Select kernel function parameters through cross-validation and the regularization parameter Specifically, the training sample set is divided into several subsets, one of which is selected as the validation set, and the remaining subsets are used as the training set; the support vector machine classification model is trained on the training set, and the model performance is evaluated on the validation set; the above steps are repeated until all subsets are used as validation sets; the optimal kernel function parameter σ and regularization parameter C are selected according to the F1 value performance index of each validation; ； in, Represents precision, Recall represents recall; The fine classification module is used to perform fine classification of soil types based on the preliminary classification results according to the geographically weighted regression algorithm to obtain fine classification results; Step S31, construct a weight matrix, for each pixel , construct the weight matrix , the elements of the weight matrix Represents pixel Pixel Weight: ； in, It's a pixel and The distance between is the bandwidth parameter; Step S32, calculate the regression coefficient for each pixel , using the weighted least squares method to calculate the regression coefficients : ； in, is the feature matrix, is the preliminary classification result; Step S33, correct the classification results, using the regression coefficient Correct the preliminary classification results to obtain refined classification results ; The type map updating module is used to update the soil type map according to the refined classification result; The steps to update soil type maps include: data integration and mapping, soil type map compilation, verification and evaluation, and results publication and application; Integrate the refined classification results with the corresponding geographic location information to form a complete soil type dataset; use the geographic information system to map the soil type dataset to the geographic space of the target area, convert the soil type labels into attribute fields in the GIS layer, and assign corresponding colors or symbols to distinguish different soil types; If the target area already has an old soil type map, the old soil type map should be overlaid with the refined classification results for analysis. The specific method is: check whether the coordinate systems of all layers are consistent; if not, use GIS software to convert the coordinates; convert all layers to a format supported by GIS software; and overlay the refined classification result layers with the old soil type map one by one through the overlay function of GIS software.