A method, device and computer for alluvial fan extraction

Abstract

The present invention provides a method, apparatus, and computer for alluvial fan extraction. The method includes: obtaining at least one historical sample data, processing the at least one historical sample data to obtain a binary classification map; constructing a training sample set according to the binary classification map; training a preset sample model using the training sample set to obtain a trained model; and inputting at least one sample data to be extracted into the trained model for processing to obtain a judgment result on whether the sample data to be extracted is an alluvial fan. The present invention can accurately classify alluvial fans in plateau areas without the need for a large amount of alluvial fan sample data.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent recognition of remote sensing images, and particularly to a method, device and computer for extracting alluvial fans. Background Art

[0002] An alluvial fan is a fan-shaped accumulation body at the exit of a river. It is distributed in different climatic and tectonic environments around the world. In the Qinghai-Tibet Plateau region, alluvial fans are mostly distributed in front of mountains. Earth and stones in the mountains gush out of the mountain pass along with rivers and floods. As the water flow slows down, earth, stones and sand gradually deposit, forming a nearly fan-shaped form. The alluvial fan has good soil and water conditions, is suitable for planting and grazing, and is an important place for agricultural and pastoral production, with many villages distributed. However, modern alluvial fans that are still in the development stage are small in scale and scarce in vegetation. Once a disaster occurs, it will cause strong damage to human production, life and engineering construction.

[0003] At the same time, many alluvial fans have experienced relatively serious floods and caused losses. The main characteristic of flood disasters occurring on alluvial fans is that the uncertainty of the flow rate is relatively large, which needs to be paid special attention to in actual risk assessment.

[0004] Once the quantity and type of alluvial fan sediments, the terrain of the slope or the moisture condition of the fan body itself change, it may lead to disasters. For example, summer rainstorms will cause floods. Coupled with the scarce vegetation on the alluvial fan and the fast runoff speed, the sediments and debris of the alluvial fan will be transported, causing greater harm to the buildings in surrounding cities and villages. Especially when the soil reaches the saturation state due to previous precipitation, this phenomenon will be aggravated, and even the urban streets will become natural riverbeds.

[0005] Therefore, it is of great significance to clarify the location and scope of alluvial fans for risk analysis.

[0006] Currently, in China, the extraction of alluvial fans usually selects remote sensing data as the only data source, and has not fully utilized the prior knowledge of the study area, and cannot extract relevant information from the formation of alluvial fans as the supporting information for remote sensing extraction. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method, device and computer for extracting alluvial fans, which can accurately classify alluvial fans in the plateau area without the premise of a large amount of alluvial fan sample data.

[0008] To solve the above technical problems, the technical solution of the present invention is as follows:

[0009] In a first aspect, a method for extracting an alluvial fan, the method includes:

[0010] Obtain at least one historical sample data, process at least one of the historical sample data to obtain a binary classification map;

[0011] Construct a training sample set according to the binary classification map;

[0012] Use the training sample set to train a preset sample model to obtain a trained model;

[0013] Input at least one sample data to be extracted into the trained model for processing to obtain a judgment result on whether the sample data to be extracted is an alluvial fan.

[0014] Further, obtaining at least one historical sample data, processing at least one of the historical sample data to obtain a binary classification map includes:

[0015] Obtain at least one historical remote sensing image;

[0016] Take at least one historical remote sensing image as at least one historical sample data;

[0017] Select ground object targets in the historical sample data;

[0018] Mark the historical sample data after selecting the ground object targets to obtain a marked area and an unmarked area;

[0019] Perform binary processing on the marked area and the unmarked area to obtain binary classification maps corresponding to the marked area and the unmarked area respectively.

[0020] Further, constructing a training sample set according to the binary classification map includes:

[0021] Obtain at least one element feature corresponding to each of the marked area and the unmarked area, and the data value corresponding to at least one element feature in the marked area or the unmarked area;

[0022] Based on the data value and the element feature corresponding to the data value, construct a training sample set corresponding to the data value and the element feature.

[0023] Further, the element feature includes at least one of a normalized difference vegetation index, slope, aspect, and local binary pattern feature.

[0024] Further, the training sample set is a text-based structure.

[0025] Further, using the training sample set to train a preset sample model to obtain a trained model includes:

[0026] If there are M features in the training samples in the training sample set, then select at each node split a feature, and use this feature as the number of features;

[0027] Calculate the Gini coefficient for each feature and the corresponding feature value;

[0028] Select the feature and feature value corresponding to the minimum Gini coefficient to split the node until all nodes stop splitting;

[0029] Train a decision tree, and vote on the category of each training sample according to the classification results of each decision tree, and select the category with the most occurrences as the attribute of the training sample.

[0030] Furthermore, the total number of training samples in the training sample set ≥ the number of decision trees in the training model.

[0031] In a second aspect, a diluvial fan extraction device includes:

[0032] An acquisition module, configured to acquire at least one historical sample data, process the at least one historical sample data to obtain a binary classification map;

[0033] A processing module, configured to construct a training sample set according to the binary classification map; use the training sample set to train a preset sample model to obtain a training model; input at least one sample data to be extracted into the training model for processing to obtain a judgment result on whether the sample data to be extracted is a diluvial fan.

[0034] In a third aspect, a computer includes:

[0035] One or more processors;

[0036] A storage device, configured to store one or more programs, and when the one or more programs are executed by the one or more processors, enable the one or more processors to implement the method described above.

[0037] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described above is implemented.

[0038] The above solution of the present invention has at least the following beneficial effects:

[0039] With the above solution of the present invention, by using the historical sample data of diluvial fans, a region-adapted diluvial fan extraction model is established, without the need for a large amount of diluvial fan sample data, and the diluvial fans in the Qinghai-Tibet Plateau region can be accurately classified. Description of the Drawings

[0040] Figure 1It is a schematic flowchart of the alluvial fan extraction method provided by the embodiments of the present invention.

[0041] Figure 2 It is a schematic diagram of the alluvial fan extraction device provided by the embodiments of the present invention. Detailed implementation manners

[0042] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art.

[0043] As Figure 1 shown, an alluvial fan extraction method is proposed in the embodiments of the present invention. The method includes the following steps:

[0044] Step 11: Obtain at least one historical sample data, process at least one of the historical sample data to obtain a binary classification map;

[0045] Step 12: Construct a training sample set according to the binary classification map;

[0046] Step 13: Use the training sample set to train a preset sample model to obtain a trained model;

[0047] Step 14: Input at least one sample data to be extracted into the trained model for processing to obtain a judgment result on whether the sample data to be extracted is an alluvial fan.

[0048] Specifically, during geological exploration, it is necessary to classify the geological features in a certain area. In step 11, it is first necessary to obtain historical sample data so as to analyze and compare this area, and then process these historical sample data to obtain a binary classification map. This classification map divides the geological features in this area into two categories, such as alluvial fans and non-alluvial fans. This classification map is the basis for constructing the training sample set in the next step. The training sample set mentioned here refers to extracting multiple sample data from the binary classification map and labeling which category they belong to, and then using these data to train a machine learning model. The quantity and quality of these sample data will affect the accuracy and stability of the model.

[0049] In step 12, after obtaining historical sample data, obtaining a binary classification map, and constructing a training sample set, the training of the machine learning model can now begin. In step 13, the process of training the model is to adjust the preset model architecture and parameters according to the training sample set so that it can accurately classify and judge new sample data. This process requires continuous iteration and optimization until the model can achieve satisfactory accuracy and stability. In step 14, when the training of the model is completed, it can be used to judge new sample data. In this specific problem, we need to input the sample data to be extracted, that is, a piece of geological image data, and then the training model will output the judgment result of whether the image is an alluvial fan. This process will be continuously tested and optimized until the model can achieve the best performance. Therefore, by using the historical sample data of alluvial fans, a regionally adapted alluvial fan extraction model is established, which does not require a large amount of alluvial fan sample data and can accurately classify alluvial fans in the Qinghai-Tibet Plateau region.

[0050] It should be noted that the historical sample data is the original remote sensing image data. When obtaining the historical sample data, it can be obtained through the following channels: for example, the Geographic Information Administration or other local government departments: national and local governments may collect a large amount of remote sensing image data and store it in a database, so these data can be obtained by applying to these institutions; another example is remote sensing image suppliers: remote sensing image suppliers can provide high-quality remote sensing image data and usually provide a complete data subscription service to meet the different needs of users; another example is open-source data sets: some researchers and organizations will publicly release some remote sensing image data sets for everyone to use, such as the Landsat data set provided by the USGS (United States Geological Survey) and the Sentinel data set provided by Google Earth Engine.

[0051] In a preferred embodiment of the present invention, the above step 11 may include:

[0052] Step 111, obtaining at least one historical remote sensing image;

[0053] Step 112, using at least one historical remote sensing image as at least one historical sample data;

[0054] Step 113, selecting the ground object targets in the historical sample data;

[0055] Step 114, marking the historical sample data after selecting the ground object targets to obtain a marked area and an unmarked area;

[0056] Step 115: Binarize the marked area and the unmarked area to obtain binary classification maps corresponding to the marked area and the unmarked area respectively.

[0057] In this embodiment, a certain number of original remote sensing images are obtained as samples. For each image, visual interpretation is performed, that is, the alluvial fan, a ground object target, is selected and marked manually to obtain the binary classification maps of the alluvial fan area and the non-alluvial fan area. Secondly, data such as the terrain, precipitation, and remote sensing images of the alluvial fan area are preprocessed. For the remote sensing image data, after cloud removal in Google Earth Engine, it is registered with other data to the same geographic coordinates and the same projection coordinates using the projection tool of ArcGIS software, and the processed data is subjected to overlay analysis using the resampling tool of ArcGIS software.

[0058] Specifically, in step 111, at least one historical remote sensing image needs to be collected first. These remote sensing images are important data sources for constructing the training sample set and model training. Remote sensing images can be obtained through tools such as satellites and airplanes, and contain image data of multiple wavelengths and multiple perspectives of the target area. In step 112, the collected historical remote sensing images are used as historical sample data for subsequent ground object target selection and marking operations. In step 113, in the historical sample data, the ground object targets to be classified are selected to ensure that the selected ground object targets can represent the type distribution within the entire area. In step 114, the positions of these targets are marked by using artificial intelligence algorithms or manual marking methods, so that the marked area and the unmarked area can be obtained. The marking process needs to be precise and accurate to improve the accuracy of the subsequent model. In step 115, the marked area and the unmarked area are binarized to obtain binary classification maps corresponding to the marked area and the unmarked area respectively. This classification map divides the marked area and the unmarked area into two categories. The marked area is used to train the model, and the unmarked area is used for model testing and verification.

[0059] In a preferred embodiment of the present invention, the above step 12 may include:

[0060] Step 121: Obtain at least one element feature corresponding to each of the marked area and the unmarked area, and the data value corresponding to the at least one element feature in the marked area or the unmarked area;

[0061] Step 122: Based on the data values and the corresponding feature characteristics of the data values, construct a training sample set corresponding to the data values and the feature characteristics. The feature characteristics include at least one of the normalized difference vegetation index, slope, aspect, and LBP (Local Binary Pattern) feature. The training sample set is in the form of a text class structure.

[0062] Specifically, in Step 121, first, at least one corresponding feature characteristic in the marked area and the unmarked area needs to be obtained. Feature characteristics refer to the attributes or indicators related to the type of ground object target in remote sensing image data, such as the normalized difference vegetation index, slope, aspect, and LBP feature, etc. These feature values can be used to distinguish different types of ground object targets and are used as the feature of the training data in the subsequent steps. In Step 122, it is necessary to construct a training sample set corresponding to the data values and the corresponding feature characteristics of the data values. Specifically, for each ground object target, obtain the specific feature value of this ground object target in the marked area, and correspond this feature with a binary variable (that is, whether this target is a target of this type) as a training sample. These training samples constitute a data set in the form of a text class structure, which is used to train a machine learning model and can be trained using different machine learning algorithms.

[0063] In this embodiment, for the selected alluvial fan samples, select the red, green, blue, near-infrared, and two short-wave infrared bands of Sentinel-2 images, as well as the normalized difference vegetation index NDVI, slope, aspect, and local binarization features to make the sample data set. Among them, Sentinel-2 remote sensing images are currently the highest-resolution open-source remote sensing images, and different element features of the image can be extracted through different band combinations; the vegetation of alluvial fans is scarce, and the normalized difference vegetation index can reflect the vegetation density; the terrain of alluvial fans is relatively gentle, and slope and aspect are important terrain features; LBP features can reflect the texture features of alluvial fans. Therefore, the above indicators can assist in the extraction of alluvial fans and effectively express the characteristics of alluvial fans. The production of the sample data set refers to extracting the corresponding data values of each grid of each data; the form of the sample data set is txt, where each row represents the numerical values of each index corresponding to a pixel; each column represents different indexes, and the last column is Non-AF or AF, indicating whether the pixel is a non-alluvial fan or an alluvial fan, respectively.

[0064] In a preferred embodiment of the present invention, the above Step 13 may include:

[0065] Step 131: If there are M features in the training samples of the training sample set, then select features during the splitting of each node, and The number of features as the features;

[0066] Step 132, calculate the Gini coefficient for each feature and the corresponding feature value;

[0067] Step 133, select the feature and feature value corresponding to the minimum Gini coefficient to split the node until all nodes no longer split;

[0068] Step 134, train the decision tree, and the category of each training sample is voted according to the classification results of each decision tree, and the category with the most occurrences is selected as the attribute of the training sample, where the total number of training samples in the training sample set ≥ the number of decision trees in the training model.

[0069] Specifically, in step 131, a certain number of features need to be selected when splitting each node, and these features are used as the basis for decision tree classification. Specifically, samples need to be randomly selected from the training sample set. Suppose there are a total of N samples, and 100 decision trees need to be constructed. Then, the root node of each decision tree needs to randomly select a sample from these N samples. Here, a random sampling method with replacement is adopted, that is, after each sample is selected, the sample is put back into the sample set to ensure that the distribution of each sampled sample is roughly the same as that of the original data set. In steps 132 and 133, feature selection and Gini coefficient calculation need to be performed on each node. In each node, a certain number of random features (usually sqrt(total number of features)) need to be selected, and the Gini coefficient is calculated for each value under each feature. The Gini coefficient can measure the importance of a feature or its feature value. The smaller the value, the more effectively the feature or feature value can classify the target. Therefore, it can be used to select the best classification basis according to the Gini coefficients of different features or feature values.

[0070] In step 134, the decision tree needs to be recursively constructed until all nodes can no longer split. Specifically, in each node, according to the selected features and the Gini coefficient calculation results, the best feature value is selected as the splitting basis for the node, the node is divided into two child nodes, and recursive operations are performed on each child node until it can no longer be split, that is, reaching the leaf node. By voting on the classification results of multiple decision trees, the final classification result is obtained. For a new unclassified data, each decision tree in the model will give a predicted classification, and the final classification result is the mode of the predicted classifications of all decision trees. In this way, the overfitting risk of a single decision tree can be reduced, and the stability and classification accuracy of the model can be improved.

[0071] As Figure 2 shown, an alluvial fan extraction device 20 according to an embodiment of the present invention further includes:

[0072] An acquisition module 21, configured to acquire at least one piece of historical sample data, process the at least one piece of historical sample data, and obtain a binary classification map;

[0073] A processing module 22, configured to construct a training sample set according to the binary classification map; use the training sample set to train a preset sample model to obtain a trained model; input at least one sample data to be extracted into the trained model for processing, and obtain a judgment result on whether the sample data to be extracted is an alluvial fan.

[0074] Optionally, acquiring at least one piece of historical sample data, processing the at least one piece of historical sample data, and obtaining a binary classification map includes:

[0075] Acquire at least one historical remote sensing image;

[0076] Use at least one historical remote sensing image as at least one piece of historical sample data;

[0077] Select ground object targets in the historical sample data;

[0078] Mark the historical sample data after selecting the ground object targets to obtain a marked area and an unmarked area;

[0079] Perform binary processing on the marked area and the unmarked area to obtain binary classification maps corresponding to the marked area and the unmarked area respectively.

[0080] Optionally, constructing a training sample set according to the binary classification map includes:

[0081] Obtain at least one element feature corresponding to each of the marked area and the unmarked area, and the data value corresponding to the at least one element feature in the marked area or the unmarked area;

[0082] Based on the data value and the element feature corresponding to the data value, construct a training sample set with the data value corresponding to the element feature.

[0083] Optionally, the element feature includes at least one of a normalized difference vegetation index, a slope, an aspect, and a local binary pattern feature.

[0084] Optionally, the training sample set is a text class structure.

[0085] Optionally, using the training sample set to train a preset sample model to obtain a trained model includes:

[0086] If there are M features in the training samples in the training sample set, then when splitting each node, select features, and use features as the number of features;

[0087] Calculate the Gini coefficient for each feature and the corresponding eigenvalue of that feature;

[0088] Select the feature and eigenvalue corresponding to the minimum Gini coefficient to split the node until all nodes stop splitting;

[0089] Train a decision tree, and vote on the category of each training sample according to the classification results of each decision tree, and select the category with the most occurrences as the attribute of the training sample.

[0090] Optionally, the total number of training samples in the training sample set ≥ the number of decision trees in the training model.

[0091] It should be noted that this device corresponds to the above method, and all implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0092] An embodiment of the present invention also provides a flight control computer for a launch vehicle, including: a processor and a memory storing a computer program. When the computer program is run by the processor, it executes the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0093] An embodiment of the present invention also provides a computer-readable storage medium storing instructions. When the instructions are run on a computer, the computer is caused to execute the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0094] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0095] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0096] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling, or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the devices or units can be in electrical, mechanical, or other forms.

[0097] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0098] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0099] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or this part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0100] In addition, it should be noted that in the device and method of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations shall be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed in chronological order according to the described order, but it is not necessary to be executed in chronological order. Certain steps can be executed in parallel or independently of each other. For those of ordinary skill in the art, it is possible to understand all or any steps or components of the method and device of the present invention, and they can be implemented in any computing device (including processors, storage media, etc.) or a network of computing devices in the form of hardware, firmware, software, or a combination thereof, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0101] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the object of the present invention can also be achieved only by providing a program product containing program code for implementing the method or device. That is to say, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be noted that in the device and method of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations shall be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed in chronological order according to the described order, but it is not necessary to be executed in chronological order. Certain steps can be executed in parallel or independently of each other.

[0102] The above are the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

[0103] The above are the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for extracting alluvial fans, characterized in that, the method includes: Obtaining at least one historical sample data, processing at least one of the historical sample data to obtain a binary classification map; Constructing a training sample set according to the binary classification map; Using the training sample set to train a preset sample model to obtain a trained model; Inputting at least one sample data to be extracted into the trained model for processing to obtain a judgment result on whether the sample data to be extracted is an alluvial fan; Among them, obtaining at least one historical sample data, processing at least one of the historical sample data to obtain a binary classification map, includes: Obtaining at least one historical remote sensing image; Taking at least one historical remote sensing image as at least one historical sample data; Selecting ground object targets in the historical sample data; Marking the historical sample data after selecting the ground object targets to obtain a marked area and an unmarked area; Performing binary processing on the marked area and the unmarked area to obtain binary classification maps corresponding to the marked area and the unmarked area respectively; Among them, constructing a training sample set according to the binary classification map includes: Obtaining at least one feature characteristic corresponding to each of the marked area and the unmarked area, and the data value corresponding to at least one feature characteristic in the marked area or the unmarked area; Based on the data value and the feature characteristic corresponding to the data value, constructing a training sample set corresponding to the data value and the feature characteristic; Among them, the feature characteristic includes at least one of a normalized difference vegetation index, slope, aspect, and local binary pattern feature; Among them, the training sample set is a text type structure; Among them, using the training sample set to train a preset sample model to obtain a trained model includes: If the training samples in the training sample set have M features, then when splitting each node, select features, and use features as the number of features; Calculating the Gini coefficient for each feature and the feature value corresponding to the feature; Selecting the feature and the feature value corresponding to the minimum Gini coefficient to split the node until all nodes no longer split; Training a decision tree, and voting on the category of each training sample according to the classification result of each decision tree, and selecting the category with the most occurrences as the attribute of the training sample; Among them, the total number of training samples in the training sample set ≥ the number of decision trees in the trained model.

2. An alluvial fan extraction device, characterized in that, it includes: An acquisition module for obtaining at least one historical sample data, processing at least one of the historical sample data to obtain a binary classification map; A processing module for constructing a training sample set according to the binary classification map; Using the training sample set to train a preset sample model to obtain a trained model; inputting at least one sample data to be extracted into the trained model for processing to obtain a judgment result on whether the sample data to be extracted is an alluvial fan; Among them, obtaining at least one historical sample data, processing at least one of the historical sample data to obtain a binary classification map, includes: Obtaining at least one historical remote sensing image; Taking at least one historical remote sensing image as at least one historical sample data; Selecting ground object targets in the historical sample data; Mark the historical sample data after selecting the ground object targets to obtain the marked area and the unmarked area; Perform binarization processing on the marked area and the unmarked area to obtain binary classification maps corresponding to the marked area and the unmarked area respectively; Among them, constructing a training sample set according to the binary classification map includes: Obtain at least one feature characteristic corresponding to each of the marked area and the unmarked area, and the data value corresponding to at least one feature characteristic in the marked area or the unmarked area; Based on the data value and the feature characteristic corresponding to the data value, construct a training sample set with the data value corresponding to the feature characteristic; Among them, the feature characteristic includes at least one of a normalized vegetation index, slope, aspect, and local binarization feature; Among them, the training sample set is a text class structure; Among them, using the training sample set to train a preset sample model to obtain a training model includes: If the training samples in the training sample set have M features, then when splitting each node, select features, and use features as the number of features; Calculate the Gini coefficient for each feature and the feature value corresponding to the feature; Select the feature and feature value corresponding to the minimum Gini coefficient to split the node until all nodes no longer split; Train a decision tree, and the category of each training sample votes according to the classification result of each decision tree, and select the category with the most occurrences as the attribute of the training sample; Among them, the total number of training samples in the training sample set ≥ the number of decision trees in the training model.

3. A computer, characterized in that, comprising: One or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, enabling the one or more processors to implement the method according to claim 1.

4. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program, and when the program is executed by a processor, it implements the method according to claim 1.

Images