Moon DEM upscaling method and system based on expert knowledge and random forest model

Abstract

The invention provides a moon DEM upscaling method and system based on expert knowledge and a random forest model, and belongs to the technical field of digital processing. Marking different types of landform areas under the initial resolution (high resolution) in the initial moon DEM data by an expert, and quantifying the marked areas to generate second raster data; the second raster data comprises labels of different types of landforms under the target resolution, so that expert knowledge is quantified into the labels under the target resolution. On the basis, training sample data sets corresponding to different types of landforms are constructed in combination with initial lunar DEM data, and the feature importance of the different training sample data sets output by the random forest model is multiplied by the feature vectors of the training samples, so that the feature importance of the different training sample data sets is obtained. The expert knowledge is transmitted from the initial lunar DEM data to the DEM data after upscaling. According to the scheme, the practicability of the lunar DEM upscaling method in lunar appearance multi-scale mapping can be improved.

Description

Technical Field

[0001] The present application relates to the field of digital processing technology, and in particular to a lunar DEM upscaling method and system based on expert knowledge and a random forest model. Background Art

[0002] The study of lunar surface geomorphology is the scientific basis for understanding the formation and evolution of the moon. At present, most of the research on lunar landforms uses lunar DEM data. In the study of lunar geomorphology, the scale (spatial resolution) issue is particularly important. At different scientific scales, the lunar landform types are also different, which has a very important impact on the results of subsequent research. In other words, different research results may be obtained based on lunar DEM data of different resolutions.

[0003] From the perspective of data availability, the currently available DEM data include: lunar topography data with a resolution of 100m / pixel generated based on the wide-angle camera data WAC carried on the US lunar orbiter LRO, and higher-resolution DEM data generated based on data obtained by the terrain camera TC (Terrain Camera) carried on the Japanese SELENE satellite. However, the available resolutions of the currently available lunar DEM data are very limited.

[0004] In order to enrich the study of lunar features, it is necessary to solve the problem of how to obtain lunar DEM data of different scales (spatial resolution) from the existing limited data. One solution is to convert the scale of the data, for example, using upscaling technology to obtain lunar DEM data of different scales. Upscaling refers to using a certain method to obtain the corresponding low-resolution data from high-resolution data, such as obtaining a 100m resolution lunar DEM from a 10m resolution lunar DEM.

[0005] The existing technology also provides some methods for upscaling Earth DEM data, such as regional averaging method, bilinear interpolation method, etc., but these methods cannot be directly applied to the upscaling of lunar DEM data. The reason is that these methods often ignore the complexity of the lunar surface, and the mapping established between low-resolution data and high-resolution data often cannot reflect the complex nonlinear relationship between the two. In addition, traditional DEM data will cause the loss of key topographic information (such as the boundaries of lunar craters, central peaks, etc.) during the scale conversion process, especially the upscaling process, and cannot meet the needs of practical applications. For example, it causes insufficient accuracy in the route planning of detection equipment (such as lunar unmanned vehicles, etc.).

[0006] Therefore, it is necessary to provide an improved technical solution to address the above-mentioned deficiencies in the prior art. Summary of the invention

[0007] The purpose of this application is to provide a lunar DEM upscaling method and system based on expert knowledge and random forest model to solve or alleviate the problems existing in the above-mentioned prior art.

[0008] In order to achieve the above objectives, this application provides the following technical solutions:

[0009] In the first aspect, the present application provides a lunar DEM upscaling method based on expert knowledge and a random forest model, including:

[0010] Experts mark different types of topographic areas at the initial resolution in the initial lunar DEM data, and quantify the marked areas to generate the second raster data;

[0011] The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data includes labels of different types of landforms at the target resolution;

[0012] Based on the second raster data and the initial lunar DEM data, construct training sample data sets corresponding to different types of landforms;

[0013] According to the training sample data set, a random forest model corresponding to different types of terrain and landforms is constructed, and the random forest model is trained until convergence, and the feature importance corresponding to the feature vector of each training sample data set is output;

[0014] The feature importance is multiplied with the feature vector of each training sample data set and post-processed to obtain the upscaling result of the lunar DEM at the target resolution.

[0015] In conjunction with the first aspect, in some possible implementations, before the experts mark different types of topographic regions at the initial resolution in the initial lunar DEM data, the method further includes:

[0016] According to the scope of the study area, the lunar DEM data at the initial resolution is clipped to obtain the initial lunar DEM data.

[0017] In conjunction with the first aspect, in some possible implementations, the experts mark different types of topographic areas at the initial resolution in the initial lunar DEM data, including:

[0018] The initial resolution was marked in the initial lunar DEM data, and the vector ranges of different types of topographic areas were marked by experts using GIS software;

[0019] All pixels of the initial lunar DEM data covered by the vector range are assigned a value of 1, and the remaining pixels are assigned a value of 0, to obtain different types of topographic areas at the initial resolution, and record them as the first raster data.

[0020] In combination with the first aspect, in some possible implementations, the marked area is quantized to generate second raster data in the following steps:

[0021] Generate unvalued raster data according to a target resolution based on a range covered by the first raster data;

[0022] All pixels in the unvalued raster data are traversed, and labels are assigned to each pixel to generate second raster data containing labels of different types of landforms.

[0023] In combination with the first aspect, in some possible implementations, the different types of topographical and geomorphic regions include: an inner area of ​​a lunar impact crater, an edge area of ​​a lunar impact crater, and other topographical and geomorphic regions;

[0024] Correspondingly, all pixels in the unvalued raster data are traversed, and labels are assigned to each pixel to generate second raster data containing labels of different types of landforms, specifically:

[0025] Recording the currently traversed pixel in the unvalued raster data as a first pixel, and determining a second pixel set corresponding to the first pixel according to a spatial position correspondence relationship between the unvalued raster data and the first raster data;

[0026] The second pixel set is a plurality of pixels corresponding to the first pixel in the first raster data;

[0027] If the values ​​of all pixels in the second pixel set are 1, the first pixel is assigned as the label of the inner area of ​​the lunar impact crater;

[0028] If the values ​​of some pixels in the second pixel set are all 1, the label of the first pixel is marked as the label of the edge area of ​​the lunar impact crater;

[0029] If the values ​​of all pixels in the second pixel set are not 1, the label of the first pixel is marked as the label of other topographical areas.

[0030] In combination with the first aspect, based on the second raster data and the initial lunar DEM data, training sample data sets corresponding to different types of topography are constructed, specifically:

[0031] Dividing the second raster data into a plurality of raster data subsets according to different types of landforms;

[0032] Traversing each pixel in each of the raster data subsets, extracting values ​​of multiple pixels corresponding to the currently traversed pixel from the initial lunar DEM data to form a feature vector;

[0033] The feature vector is combined with the label of the currently traversed pixel to obtain a training sample data set corresponding to each of the raster data subsets.

[0034] In combination with the first aspect, in some possible implementations, the feature importance is multiplied by the feature vector of each training sample data set and post-processed to obtain the upscaling result of the lunar DEM at the target resolution, including:

[0035] Normalizing the feature importance to obtain normalized feature importance;

[0036] The normalized feature importance is multiplied by the feature vector of each training sample data set to obtain a DEM value set corresponding to different types of landforms after upscaling;

[0037] The DEM value sets corresponding to the different types of landforms after upscaling are shaped to generate the upscaling results of the lunar DEM at the target resolution.

[0038] In combination with the first aspect, in some possible implementations, the DEM value sets corresponding to the upscaled different types of landforms are shaped to generate the upscaled lunar DEM result at the target resolution, including:

[0039] Rearranging the DEM value sets corresponding to the different types of landforms after the upscaling to obtain restored raster data;

[0040] According to different types of landforms, different raster data subsets are used to perform mask operations on the restored raster data, and the mask results of different types of landforms are superimposed to obtain the upscaling results of the lunar DEM at the target resolution.

[0041] In the second aspect, this embodiment provides a lunar DEM upscaling system based on expert knowledge and a random forest model, including:

[0042] a knowledge quantification unit configured to have an expert mark different types of topographic regions at an initial resolution in the initial lunar DEM data, and quantify the expert knowledge to generate second raster data;

[0043] The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data includes labels of different types of landforms at the target resolution;

[0044] A sample generating unit, configured to construct training sample data sets corresponding to different types of landforms based on the second raster data and the initial lunar DEM data;

[0045] A model training and output unit, configured to construct a random forest model corresponding to different types of terrain and landforms according to the training sample data set, train the random forest model until convergence, and output the feature importance corresponding to the feature vector of each training sample data set;

[0046] The result generating unit is configured to multiply the feature importance with the feature vector of each training sample data set correspondingly and perform post-processing to obtain the upscaling result of the lunar DEM at the target resolution.

[0047] In a third aspect, this embodiment provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the lunar DEM upscaling method described in any of the above embodiments when executing the computer program.

[0048] The embodiment of the present application provides a lunar DEM upscaling method and system based on expert knowledge and random forest model, in which experts mark different types of topographic areas at the initial resolution (high resolution) in the initial lunar DEM data, and quantify the marked areas to generate second raster data; the second raster data contains labels of different types of topography at the target resolution (low resolution), which is essentially to quantify the expert knowledge into labels at the target resolution. On this basis, the initial lunar DEM data is combined to construct training sample data sets corresponding to different types of topography, and the feature importance of different training sample data sets output by the random forest model is multiplied by the feature importance and the feature vector of the training sample to realize the transfer of expert knowledge from the initial lunar DEM data to the upscaled DEM data. In short, this technical solution quantifies expert knowledge into labels at the target resolution and uses the random forest model to transfer expert knowledge, thereby establishing a nonlinear mapping relationship between the lunar initial resolution DEM data and the target resolution DEM data. This allows the key topographic information expressed by the lunar DEM data at the initial resolution to be retained at the target resolution. This not only alleviates the data acquisition problem of lunar DEMs at different scales, but also ensures that the upscaled lunar DEM data can meet the needs of actual applications, thereby improving the practicality and applicability of the lunar DEM upscaling method in multi-scale lunar mapping. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a schematic diagram of key topographic information in lunar mapping.

[0050] Figure 2 A schematic flowchart of a lunar DEM upscaling method based on expert knowledge and a random forest model is provided for some embodiments of the present application.

[0051] Figure 3 Some embodiments of the present application provide a technical logic block diagram of a lunar DEM upscaling method based on expert knowledge and a random forest model.

[0052] Figure 4 A schematic structural diagram of a lunar DEM upscaling system based on expert knowledge and a random forest model provided for some embodiments of the present application.

[0053] Figure 5 A schematic diagram of the structure of an electronic device provided for some embodiments of the present application.

[0054] Description of reference numerals:

[0055] 1- crater rim, 2- crater wall, 3- crater bottom, 4- central peak, 5- lunar crust. DETAILED DESCRIPTION

[0056] In order to facilitate understanding of the technical solution of the present disclosure, Figure 1 The key topographic and geomorphic information in lunar mapping is explained.

[0057] Lunar impact craters are the most common and significant geomorphic units and geotectonic landmarks on the lunar surface. They are numerous and varied in shape, presenting a circular pit structure of varying sizes and uneven concentration. The typical profile of an impact crater is as follows: Figure 1 As shown. Figure 1It can be seen that the lunar impact crater includes: rim 1 (rim of the impact crater), crater wall 2, crater bottom 3, and central peak 4. Among them, rim 1 refers to the more prominent edge part around the impact crater, which is often circular or elliptical and slightly higher than the surrounding terrain. Crater wall 2 refers to the steep inner wall part extending from rim 1 to bottom 3 inside the impact crater. Crater bottom 3 refers to the lowest flat or slightly undulating area inside the impact crater, located between crater wall 2 and the central mountain or directly filling the entire impact crater bottom 3. Central peak 4 refers to the raised ridge or peak formed in the central area of ​​a larger impact crater (usually with a diameter of more than 15-20 kilometers). The impact crater is located on the lunar crust 5, which not only affects the appearance of the lunar surface, but also has an important impact on lunar exploration. At present, lunar exploration is still unmanned deep space exploration. Before exploration, it is often necessary to map the lunar landscape of the exploration area to plan the forward route of the exploration equipment (such as lunar unmanned vehicles, etc.). If the key topographic information of the impact crater cannot be retained in the lunar landscape mapping, it may cause the exploration equipment to be unable to avoid obstacles. Moreover, the shape of the impact crater not only affects the obstacle avoidance of the detection equipment, but also affects the path selection and driving efficiency. That is to say, if the location and shape of the impact crater cannot be accurately reflected in the lunar map, path planning may need to choose a more circuitous route, increasing the detection time and energy consumption, which may result in the detection device failing to accurately reach the target location, affecting the quality of the lunar exploration data and the detection results.

[0058] As described in the background technology section, in the prior art, the available lunar DEM data is very limited. For example, the lunar DEM data that is relatively easy to obtain includes lunar DEMs with a resolution of 10 meters and 100 meters. The lunar DEMs with other resolutions have not yet formed mature data products and cannot meet the needs of multi-scale mapping of the lunar landscape. When the lunar DEM data is upscaled using the traditional earth DEM upscaling method, it is easy to cause the loss of key information such as the boundary and central peak of the lunar impact crater, resulting in the impact crater after upscaling often showing changes in shape, such as the disappearance of the central peak and the confusion between the crater wall and the crater edge.

[0059] In view of this, the present disclosure provides a lunar DEM upscaling method and system based on expert knowledge and random forest model. The method fully considers the complexity of the lunar surface topography and geomorphology, and establishes a nonlinear mapping relationship between high-resolution DEM data and low-resolution DEM data by combining expert knowledge that is difficult to quantify with a machine learning algorithm (random forest algorithm), so as to improve the mapping accuracy of the lunar DEM data after upscaling.

[0060] The embodiments of the present application are described below in conjunction with the accompanying drawings.

[0061] The terms "first", "second", "third" and "fourth" etc. in the specification and claims of the present application and the drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0062] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0063] "Multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the related objects are in an "or" relationship.

[0064] Embodiment 1:

[0065] This embodiment provides a lunar DEM upscaling method based on expert knowledge and a random forest model, which can be executed by an electronic device, such as Figure 2 As shown, the method comprises the following steps:

[0066] Step S101: An expert marks different types of topographic regions at an initial resolution in the initial lunar DEM data (hereinafter referred to as grid A), and quantifies the marked regions to generate second grid data (hereinafter referred to as grid C);

[0067] The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data contains labels of different types of landforms at the target resolution;

[0068] Step S102: constructing training sample data sets corresponding to different types of topography based on the second raster data and the initial lunar DEM data;

[0069] Step S103: construct random forest models corresponding to different types of terrain and landforms according to the training sample data set, train the random forest model until convergence, and output the feature importance corresponding to the feature vector of each training sample data set;

[0070] Step S104: multiply the feature importance with the feature vector of each training sample data set and perform post-processing to obtain the upscaling result of the lunar DEM at the target resolution.

[0071] In this embodiment, expert knowledge refers to the knowledge and insights possessed by experts in the field of lunar mapping who have rich experience, in-depth understanding and high-level skills. Expert knowledge can be used to determine the key terrain and geomorphic information that is indispensable in lunar mapping.

[0072] In this embodiment, the initial lunar DEM data refers to the lunar DEM data with initial resolution that can be obtained. In the application scenario of upscaling, the initial resolution refers to high resolution, and the resolution after upscaling (target resolution) is low resolution. The traditional method is used to generate low-resolution lunar DEM data from high-resolution lunar DEM data. Cartographic synthesis is usually used to simplify, abstract and refine the topography to obtain the lunar mapping result at low resolution (target resolution).

[0073] On the lunar surface, different types of landforms can include impact craters, lunar crust, lunar seas, lunar highlands, etc.

[0074] Furthermore, in this embodiment, according to the purpose of mapping, different types of landforms can be specifically the detailed landforms of the impact crater, including the inner area of ​​the lunar impact crater, the edge area of ​​the lunar impact crater, and other landform areas.

[0075] The inner area of ​​the lunar crater can include all areas within the edge of the crater. The edge area of ​​the lunar crater refers to the edge of the crater. Other topographic areas are all areas except the inner and edge of the crater.

[0076] In this embodiment, based on expert knowledge, different types of topographic and geomorphic element areas are identified and marked on the lunar DEM of the initial resolution.

[0077] Here, "labeling" refers to the process by which experts use their expertise and tools (such as GIS software) to classify, identify and mark different types of topographic areas in the lunar digital elevation model (DEM) data. In this process, experts first analyze the initial lunar DEM data to determine the distribution range and boundaries of different topographic features; then divide the topography into several types, such as the interior of the crater, the edge, and other topography; on this basis, determine the specific geographical location of each type of topography and mark it in the form of vector polygons.

[0078] Quantify the marked area, including: scale conversion and assigning corresponding labels or classification values ​​to each area. For example, the pixel at the edge of the impact crater is marked as 1, the pixel inside the impact crater is marked as 2, and other landform types are marked as 3. Scale conversion refers to the upscaling operation, that is, aggregating the quantization results at the initial resolution to generate raster data at the target resolution. For the convenience of description, it is called the second raster data.

[0079] Among them, the target resolution is the resolution after upscaling. The increase in scale will lead to a decrease in spatial resolution. Therefore, the target resolution is lower than the initial resolution. For example, the initial resolution is 10 meters, and the target resolution can be 25 meters, 50 meters or 100 meters, or other resolutions. This embodiment does not limit the values ​​of the initial resolution and the target resolution.

[0080] Here, it should be noted that in geographic information data, resolution (spatial resolution) reflects the data's ability to describe geographic spatial details. The higher the resolution, the smaller the surface area covered by the pixel, indicating more detailed geographic information. For example, DEM data with a resolution of 1 meter means that each pixel covers an area of ​​1×1 meter on the surface, while DEM data with a resolution of 10 meters means that each pixel covers an area of ​​10×10 meters on the surface. Relatively speaking, 1 meter resolution is high resolution and 10 resolution is low resolution.

[0081] In this embodiment, the resolution after upscaling, i.e., the target resolution, needs to be determined by the target determination module. The target resolution is determined according to the mapping scale, which is determined according to the research needs. At the same time, the landform mapping area, i.e., the study area, needs to be determined according to the research needs. The specific steps can be as follows:

[0082] First, determine the scope of the study area: determine the lunar area that needs to be mapped based on the research objectives. For example, the study area can be based on a certain geographical area of ​​the moon (such as a specific longitude and latitude range) or a research mission (such as lunar rover path planning). Existing maps or coordinates (for example, specific longitude and latitude or patch data) can be used to define the study area.

[0083] Then, determine the cartographic scale: The cartographic scale is used to reflect the ratio of the actual surface to the map. The choice of scale depends on the purpose of the study, data resolution, and application requirements of the map.

[0084] Next, determine the target resolution based on the mapping scale: There is a direct relationship between scale and resolution. Smaller scales (i.e., covering a larger surface area) generally require lower resolutions, while larger scales require higher resolutions to show more details. Resolution is the actual ground distance represented by each pixel, usually expressed in meters per pixel. The larger the scale, the finer the resolution, and the clearer the landform features that can be displayed. Based on the scale relationship, the target resolution can be estimated using the following formula:

[0085]

[0086] For example, if the mapping requirement is 1:1 million, that is, the target scale is 1:1 million, then according to the formula, the upscaled DEM data is estimated to be 100m (based on the human eye's highest visual resolution of 0.1mm).

[0087] After determining the target resolution, the second raster data can be generated according to the target resolution. In this embodiment, in addition to the change in resolution, the second raster data also includes labels of different types of landforms at the target resolution, that is, by labeling the landforms and quantifying them into labels at the target resolution, the expert's classification experience and labeling results of the landforms are converted into a digital and standardized form, and mapped to the pixels of the raster data to form the second raster data, so that the expert knowledge is converted into a form that is computable and scalable, laying the foundation for the subsequent transfer and use of expert knowledge.

[0088] Step S102 aims to combine the second raster data containing expert knowledge with the initial lunar DEM data to generate training samples for the random forest model. Since the second raster data contains labels of different types of landforms at the target resolution, the second raster data can be used to provide label values ​​for the training samples, and features can be extracted from the initial lunar DEM data to form corresponding feature vectors, and then the labels and feature vectors are combined to generate a training sample data set.

[0089] What is constructed in this embodiment is a training sample data set corresponding to different types of landforms and geomorphs, that is, the training sample data set is organized according to different types of landforms and geomorphs, and each type of landforms and geomorphs corresponds to a training sample data set. For example, the training sample data set may include: a training sample data set for the inner area of ​​the lunar impact crater, a training sample data set for the edge area of ​​the lunar impact crater, and a training sample data set for other landforms and geomorphic areas. The training sample data sets are organized according to different types of landforms and geomorphs. The independent data sets allow each type of landforms and geomorphs to be trained separately, which facilitates the model's deep learning of single-category features, helps to adjust the weight of each landform type during the training process, avoids the problem of category imbalance in the training data, and improves the pertinence of model training. At the same time, such a data organization form is also convenient for docking with expert knowledge. Experts can easily review and update training sample data sets of a single category without having to review and reconstruct training sample data sets of all categories.

[0090] In this embodiment, step S103 aims to build independent models for different types of landforms, train the model until convergence, ensure that the model can fully learn the characteristics of each landform, and output the feature importance corresponding to each training sample data set to quantify the contribution of each input feature to the prediction result. Since the input features are extracted from the initial lunar DEM data and the labels are generated by expert knowledge, the feature importance can reflect the results of the expert knowledge quantified and learned by the model.

[0091] In step S104, the feature importance is multiplied by the feature vector of each training sample data set to migrate the parameters learned based on expert knowledge from the initial DEM data to the DEM value set at the target resolution. Through post-processing operations, the DEM value set is reorganized into a raster data form with spatial location, thereby obtaining the upscaling result of the lunar DEM at the target resolution.

[0092] It should be noted that, in this embodiment, multiplying the feature importance by the feature vector of each training sample data set correspondingly refers to finding the inner product between the feature importance and the feature vector, that is, multiplying and summing the corresponding elements of the two vectors.

[0093] Since feature importance is obtained based on expert knowledge and can reflect the contribution of each feature to model prediction, when the feature importance is inner-producted with the feature vector, it actually measures the relationship between the importance of each feature in the current sample and the feature value, thereby quantifying the contribution of each feature to the model prediction results. In this way, expert knowledge can be quantitatively transferred from the initial resolution to each pixel of the target resolution, so that the model can accurately estimate the pixel value after upscaling.

[0094] In summary, the method provided in this embodiment, according to the purpose of mapping, annotates the initial DEM data, quantifies expert knowledge, constructs a training sample data set and trains the random forest model of each type of landform respectively, realizes the migration of parameters after expert knowledge learning by correspondingly multiplying the feature importance with the feature vector of each training sample data set, generates a set of upscaled DEM values, and then uses the post-processing module to generate the upscaled result of the lunar DEM at the target resolution. The above process ensures the invariance of expert knowledge during the upscaling process, and provides scientific and effective DEM data for lunar landscape research and mapping.

[0095] In an optional embodiment, before experts mark different types of terrain and landform areas at the initial resolution in the initial lunar DEM data, the method also includes: cropping the lunar DEM data at the initial resolution according to the scope of the study area to obtain the initial lunar DEM data.

[0096] In this embodiment, the clipping of the lunar DEM data is completed through the data acquisition and preprocessing module.

[0097] When acquiring lunar DEM data, in order to preserve high-precision landform information as much as possible in lunar mapping, higher-resolution DEM data can be selected as the initial lunar DEM data. For example, digital terrain data with a resolution of 10 meters / pixel generated by SELENE TC (Terrain Camera) data can be used as the initial lunar DEM data. Then, according to the scope of the study area, use geographic information system (GIS) software (such as QGIS, ArcGIS) to load the initial lunar DEM data into the software in the form of raster data. Then use the clipping tool in the GIS software to clip the loaded DEM data according to the boundary of the study area. After clipping is completed, check whether the output DEM data meets expectations, ensure that the clipped data correctly covers the study area, and no important terrain information is lost.

[0098] In this embodiment, by cropping the lunar DEM data at the initial resolution, the focus of data processing is ensured, making subsequent data processing and model training more accurate and efficient.

[0099] In an optional embodiment, different types of topographic regions at an initial resolution are marked by experts in the initial lunar DEM data, including: marking the vector ranges of different types of topographic regions at an initial resolution in the initial lunar DEM data by experts using GIS software; assigning a value of 1 to all pixels of the initial lunar DEM data covered by the vector range and a value of 0 to the remaining pixels, thereby obtaining different types of topographic regions at the initial resolution, and recording them as the first raster data (hereinafter also referred to as: raster B).

[0100] In the above embodiment, the labeling method of first labeling the vector range and then rasterizing is adopted. The advantage is that experts can flexibly adjust the boundaries of landform types according to actual conditions, which is more intuitive and accurate than operating directly in the raster, which is conducive to improving the flexibility of labeling, and the vector data can accurately represent the boundaries of the landform area, and the boundaries of the landform area can be accurately labeled by GIS software to ensure the high accuracy of the data. By assigning 1 to all pixels of the initial lunar DEM data covered by the vector range, a clear geographical range is provided for the label of the upscaled lunar DEM data, which can ensure that when the raster data at the target resolution is subsequently generated, the corresponding label values ​​are correctly and quickly assigned to the corresponding pixels at the target resolution using multiple pixel values ​​at the initial resolution.

[0101] In an optional embodiment, the marked area is quantified to generate second grid data, and the steps are as follows:

[0102] Step S1011 : Generate unvalued raster data according to the range covered by the first raster data and the target resolution.

[0103] It should be noted that, here, the unassigned raster data means that the resolution of the raster data has been upscaled from the initial resolution to the target resolution, but it is uncertain which terrain type each pixel in the raster data should belong to, and it needs to be determined through the label assignment process of step S1012.

[0104] In this embodiment, the unvalued raster data is generated based on the range covered by the first raster data. That is to say, the geographical range covered by the unvalued raster data is the same as the range covered by the first raster data. In other words, pixels outside the range covered by the first raster data do not contain expert knowledge and do not participate in the subsequent label assignment process.

[0105] Step S1012: traverse all pixels in the unvalued raster data, and assign labels to each pixel to generate second raster data containing labels of different types of landforms.

[0106] In order to ensure that the result of label assignment can correctly reflect the key information of lunar appearance, on the basis of the above embodiment, a further technical solution is that different types of topographic areas include: the inner area of ​​the lunar crater, the edge area of ​​the lunar crater, and other topographic areas; correspondingly, all pixels in the unassigned raster data are traversed, and labels are assigned to each pixel to generate a second raster data containing labels of different types of topography, specifically:

[0107] Step S1012a: record the currently traversed pixel in the unvalued raster data as the first pixel, and determine the second pixel set corresponding to the first pixel based on the spatial position correspondence between the unvalued raster data and the first raster data.

[0108] The second pixel set is a plurality of pixels corresponding to the first pixel in the first raster data.

[0109] In the upscaling process, the first raster data is a high-resolution raster, and the unvalued raster data is a low-resolution raster. The spatial position relationship between the high-resolution raster and the low-resolution raster data is mainly reflected in: multiple pixels in the high-resolution raster will be "aggregated" or "merged" into a pixel of the low-resolution raster, that is, there is a positional mapping relationship between the high-resolution raster data and the low-resolution raster data. For each pixel of the raster data at low resolution, there are multiple pixels corresponding to it in the high-resolution raster data. Therefore, multiple pixels in the first raster data corresponding to any pixel of the unvalued raster data can be determined to form a second pixel set.

[0110] Step S1012b: if the values ​​of all pixels in the second pixel set are 1, the first pixel is assigned the label of the inner area of ​​the lunar impact crater;

[0111] Step S1012c: if the values ​​of some pixels in the second pixel set are all 1, the label of the first pixel is marked as the label of the edge area of ​​the lunar impact crater;

[0112] Step S1012d: If the values ​​of all pixels in the second pixel set are not 1, the label of the first pixel is marked as the label of other topographical areas.

[0113] The purpose of step S1012b to step S1012d is to determine the value of the first pixel according to the nonlinear mapping through the values ​​of all pixels in the second pixel set, that is, to determine what type of landform the lunar surface covered by the first pixel is. Specifically, if the values ​​of all pixels in the second pixel set are 1, it indicates that the lunar surface area covered by all pixels in the set is the inner area of ​​the lunar impact crater, and then the label of the first pixel is marked as the label of the inner area of ​​the lunar impact crater; if the values ​​of some pixels in the second pixel set are 1, it indicates that the area covered by the second pixel set is the edge of the impact crater, and then the label of the first pixel is marked as the label of the edge area of ​​the lunar impact crater; if the values ​​of all pixels in the second pixel set are not 1, that is, all pixels are 0, it indicates that the area covered by the second pixel set is other landform areas, and then the label of the first pixel is marked as the label of other landform areas.

[0114] Furthermore, the label values ​​of the inner area, edge area, and other landform areas of the lunar crater can be determined as needed. For example, the label of the edge area of ​​the lunar crater is 1, the label of the inner area of ​​the lunar crater is 2, and the label of other landform areas is 3. Then, the pixels in the second raster data have three possible values, namely 1, 2, and 3, and each value represents a different type of landform.

[0115] In some optional embodiments, constructing training sample data sets corresponding to different types of terrain and landforms based on the second raster data and the initial lunar DEM data may specifically include the following steps:

[0116] The second raster data is divided into a plurality of raster data subsets according to different types of landforms.

[0117] Specifically, the second raster data is divided according to the different values ​​of each pixel. For example, if the label of the edge area of ​​the lunar crater is 1, all pixels with a value of 1 are extracted to form the label data of the inner area of ​​the lunar crater, and the extracted pixels are assigned a value of 1, and all other pixels are assigned a value of 0. The result is recorded as the raster data subset C1. Similarly, all pixels with a value of 2 are extracted to form the label data of the edge area of ​​the lunar crater, and the extracted pixels are assigned a value of 1, and all other pixels are assigned a value of 0. The result is recorded as the raster data subset C2. All pixels with a value of 3 are extracted to form the label data of other topographic areas, and the extracted pixels are assigned a value of 1, and all other pixels are assigned a value of 0. The result is recorded as the raster data subset C3.

[0118] In an optional embodiment, the division of multiple raster data subsets can also be performed simultaneously with the step of label assignment, that is, in steps S1012b to S1012d, while assigning labels to various types of terrain and landforms, different types of terrain and landforms are divided into corresponding raster subsets, and in each raster subset, the pixel value = 1 is defined as the corresponding type of terrain and landform that needs to be retained, and 0 is defined as others.

[0119] Traverse each pixel in each raster data subset, extract the values ​​of multiple pixels corresponding to the currently traversed pixel from the initial lunar DEM data, and form a feature vector;

[0120] The feature vector is combined with the label of the currently traversed pixel to obtain the training sample data set corresponding to each raster data subset.

[0121] The information of a single pixel may not be sufficient to fully describe the terrain characteristics after upscaling. By extracting the values ​​of multiple pixels from the initial lunar DEM data to form a feature vector, local spatial patterns and terrain changes can be captured, thereby improving the model's discriminative ability. By combining the feature vector with the label, a training sample containing spatial information can be provided for each pixel. At the same time, each training sample contains the characteristics of the location (i.e., the values ​​of multiple pixels) and the corresponding labels (i.e., the terrain type), which is conducive to the model learning the nonlinear mapping relationship between the terrain type and its spatial characteristics, enabling the model to more accurately identify the boundaries and structures of impact craters on the lunar surface and improve the model accuracy.

[0122] Furthermore, the feature importance is multiplied with the feature vector of each training sample data set and post-processed to obtain the upscaling result of the lunar DEM at the target resolution, including:

[0123] Normalize the feature importance to obtain the normalized feature importance;

[0124] The normalized feature importance is multiplied by the feature vector of each training sample data set to obtain the DEM value set corresponding to different types of landforms after upscaling;

[0125] The upscaled DEM value sets corresponding to different types of landforms are shaped to generate the upscaled lunar DEM results at the target resolution.

[0126] Specifically, the DEM value sets corresponding to different types of landforms after upscaling are shaped to generate the upscaling results of the lunar DEM at the target resolution, including:

[0127] Rearrange the DEM value sets corresponding to different types of landforms after upscaling to obtain restored raster data;

[0128] According to different types of landforms, different raster data subsets are used to perform mask operations on the restored raster data, and the mask results of different types of landforms are superimposed to obtain the upscaling results of the lunar DEM at the target resolution.

[0129] Refer to the following Figure 3 , the technical solution provided in this embodiment is illustrated by examples.

[0130] like Figure 3 As shown, this embodiment can be implemented according to the following steps:

[0131] Step 1: Target determination module

[0132] First, determine the geomorphic mapping area and mapping scale according to the research needs, and determine the resolution after scaling according to the mapping scale. For example, if the mapping requirement is 1:100w, the DEM data after scaling should be 100m (based on the highest visual resolution of the human eye of 0.1mm).

[0133] Step 2: Data acquisition and preprocessing module

[0134] In order to preserve high-precision geomorphic information as much as possible during mapping, a higher-resolution DEM data is selected as the initial DEM data, such as digital terrain data with a resolution of 10m / pixel generated by SELENE TC (Terrain Camera) data as the initial DEM data. According to the scope of the study area, the initial DEM data is clipped to obtain the DEM data A (initial lunar DEM data) of the study area.

[0135] Step 3: Expert knowledge quantification module

[0136] According to the purpose of mapping, DEM data A is annotated to quantify expert knowledge. Specifically, (1) experts use GIS software (such as ArcGIS) to annotate the topographic and geomorphic elements that are indispensable after upscaling at the resolution of the initial DEM to obtain the corresponding vector S; (2) using DEM data A as a template, vector S is rasterized into raster B, where the regional raster value corresponding to vector S is 1 and the rest is 0; (3) raster B is aggregated and upscaled to obtain a raster set C (i.e., raster C) at the target resolution:

[0137] C={C1,C2,C3},

[0138] That is to say, grid set C is composed of multiple grid subsets, namely C1, C2, and C3. Aggregate upscaling obtains grids C1, C2, and C3 according to the following rules:

[0139] Grid C1: During the upscaling process, as long as any number of elements in each pixel of grid C and the multiple pixels corresponding to grid B are 1 (some are 1, but not all are 1), the pixel value of grid C after upscaling is 1, otherwise it is 0;

[0140] Grid C2: During the upscaling process, if all elements in the multiple pixels corresponding to each pixel of grid C and grid B are 1, then the pixel value of grid C after upscaling is 1, otherwise it is 0;

[0141] Grid C3: During the upscaling process, if no element in each pixel of grid C and the multiple pixels corresponding to grid B is 1, the pixel value of grid C after upscaling is 1, otherwise it is 0;

[0142] Step 4: Training sample construction module

[0143] Based on the data obtained in the previous step, a training sample data set is constructed. Specifically, for each pixel of grid C1, the value of grid data A in the corresponding area of ​​each pixel is extracted. Assuming that each pixel of grid C1 corresponds to 100 pixels of grid A, an X vector (x1, x2, ..., x 100 ) and a corresponding Y value, where the Y value is the label and Y is the pixel value of the grid C1, specifically 0 or 1.

[0144] Traverse each pixel of the low-resolution DEM data (grid C1) to obtain a series of X vectors and Y values ​​to form a training sample data set DATA1.

[0145] For grids C2 and C3, DATA2 and DATA3 are obtained in the same way.

[0146] Step 5: Random Forest Model

[0147] (1) Based on the training sample data set DATA1 obtained in the previous step, a random forest classification model Model-1 is constructed. The model is trained until convergence, and the importance X1 corresponding to the X vector is obtained, such as (x11, x12, ..., x1 100 ), and normalize X1 to obtain X1_α.

[0148] (2) Based on the training sample data set DATA2 obtained in the previous step, a random forest classification model Model-2 is constructed. The model is trained until convergence, and the importance X2 corresponding to the X vector is obtained, such as (x21, x22, …, x2 100 ), and normalize X2 to obtain X2_α.

[0149] (3) Based on the training sample data set DATA3 obtained in the previous step, a random forest classification model Model-3 is constructed. The model is trained until convergence, and the importance X3 corresponding to the X vector is obtained, such as (x31, x32, ..., x3 100 ), and normalize X3 to obtain X3_α.

[0150] Step 6: Parameter Migration Module

[0151] (1) For the training data set DATA1 constructed above, each sub-element (i.e., each eigenvector in the training data set DATA1) is multiplied by X1_α to obtain the upscaled DEM value set Y1;

[0152] (2) For the training data set DATA2 constructed above, each sub-element (i.e., each eigenvector in the training data set DATA2) is multiplied by X2_α to obtain the upscaled DEM value set Y2;

[0153] (3) For the training data set DATA3 constructed above, each sub-element (i.e., each eigenvector in the training data set DATA3) is multiplied by X3_α to obtain the upscaled DEM value set Y3;

[0154] Step 7: Post-processing module

[0155] Reshape Y1, Y2 and Y3 to restore them to raster data, and then use raster C1, C2 and C3 to mask the raster data obtained from Y1, Y2 and Y3 respectively, and superimpose the masked raster into a new raster to obtain the upscaled DEM data.

[0156] In this way, after the above operations, the expert knowledge is quantified, and on this basis, the upscaling operation of the DEM data is completed, maintaining the invariance of the expert knowledge in the upscaling process.

[0157] Embodiment 2:

[0158] This embodiment provides a lunar DEM upscaling system based on expert knowledge and random forest model. Figure 4 As shown, the system includes: a knowledge quantification unit 401, a sample generation unit 402, a model training and output unit 403 and a result generation unit 404. Specifically:

[0159] The knowledge quantification unit 401 is configured to have experts mark different types of topographic regions at an initial resolution in the initial lunar DEM data, and quantify the expert knowledge to generate second raster data;

[0160] The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data contains labels of different types of landforms at the target resolution;

[0161] The sample generating unit 402 is configured to construct training sample data sets corresponding to different types of landforms based on the second raster data and the initial lunar DEM data;

[0162] The model training and output unit 403 is configured to construct a random forest model corresponding to different types of terrain and landforms according to the training sample data set, train the random forest model until convergence, and output the feature importance corresponding to the feature vector of each training sample data set;

[0163] The result generating unit 404 is configured to multiply the feature importance with the feature vector of each training sample data set correspondingly and perform post-processing to obtain the upscaling result of the lunar DEM at the target resolution.

[0164] The lunar DEM upscaling system based on expert knowledge and random forest model provided in this embodiment can implement the steps and processes of the lunar DEM upscaling method based on expert knowledge and random forest model provided in any of the above embodiments, and achieve the same technical effects, which will not be repeated here one by one.

[0165] Embodiment three:

[0166] Based on the same inventive concept, this embodiment also provides an electronic device, Figure 5 is a schematic diagram of the structure of the electronic device; Figure 5 As shown, the hardware structure of the electronic device may include: a processor 501 , a communication interface 502 , a computer-readable storage medium (also called a memory) 503 and a communication bus 504 .

[0167] The processor 501 , the communication interface 502 , and the computer-readable storage medium 503 communicate with each other via a communication bus 504 .

[0168] The computer-readable storage medium 503 may be configured to store one or more programs.

[0169] Optionally, the communication interface 502 may be an interface of a communication module, such as an interface of a GSM module.

[0170] The processor 501 executes one or more programs, which implement the following steps:

[0171] Experts mark different types of topographic areas at the initial resolution in the initial lunar DEM data, and quantify the marked areas to generate the second raster data;

[0172] The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data contains labels of different types of landforms at the target resolution;

[0173] Based on the second grid data and the initial lunar DEM data, construct the corresponding training sample data sets of different types of terrain and landforms;

[0174] According to the training sample data set, a random forest model corresponding to different types of terrain and landforms is constructed, and the random forest model is trained until convergence, and the feature importance corresponding to the feature vector of each training sample data set is output;

[0175] The feature importance is multiplied with the feature vector of each training sample dataset and post-processed to obtain the upscaling result of the lunar DEM at the target resolution.

[0176] The processor 501 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0177] The electronic device of the embodiment of the present application exists in various forms, including but not limited to:

[0178] (1) Mobile communication devices: These devices are characterized by their mobile communication functions and their main purpose is to provide voice and data communications. These terminals include: smart phones (e.g., iPhone), multimedia phones, functional phones, and low-end phones.

[0179] (2) Ultra-mobile personal computer devices: These devices fall into the category of personal computers, have computing and processing capabilities, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, such as iPad.

[0180] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players (such as iPods), handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0181] (4) Server: A device that provides computing services. The server consists of a processor, hard disk, memory, system bus, etc. The server is similar to a general computer architecture, but because it needs to provide highly reliable services, it has higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.

[0182] (5) Other electronic devices with data interaction functions.

[0183] It should be pointed out that, according to the needs of implementation, the various components / steps described in the embodiments of the present application can be split into more components / steps, or two or more components / steps or partial operations of components / steps can be combined into new components / steps to achieve the purpose of the embodiments of the present application.

[0184] The above-mentioned method according to the embodiment of the present application can be implemented in hardware, firmware, or implemented as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk or magneto-optical disk), or implemented as computer code originally stored in a remote recording medium or a non-temporary machine storage medium downloaded through a network and to be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor or programmable or dedicated hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the lunar DEM upscaling method based on expert knowledge and a random forest model described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown herein.

[0185] Those of ordinary skill in the art will appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and constraints involved in the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present application.

[0186] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A lunar DEM upscaling method based on expert knowledge and random forest model, characterized in that: include: Experts mark different types of topographic areas at the initial resolution in the initial lunar DEM data, and quantify the marked areas to generate the second raster data; The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data includes labels of different types of landforms at the target resolution; Based on the second raster data and the initial lunar DEM data, construct training sample data sets corresponding to different types of landforms; According to the training sample data set, a random forest model corresponding to different types of terrain and landforms is constructed, and the random forest model is trained until convergence, and the feature importance corresponding to the feature vector of each training sample data set is output; The feature importance is multiplied with the feature vector of each training sample data set and post-processed to obtain the upscaling result of the lunar DEM at the target resolution.

2. The method according to claim 1, characterized in that Before the experts mark different types of topographic areas at the initial resolution in the initial lunar DEM data, the method further includes: According to the scope of the study area, the lunar DEM data at the initial resolution is clipped to obtain the initial lunar DEM data.

3. The method according to claim 1, characterized in that The experts marked different types of topographic areas at the initial resolution in the initial lunar DEM data, including: The initial resolution was marked in the initial lunar DEM data, and the vector ranges of different types of topographic areas were marked by experts using GIS software; All pixels of the initial lunar DEM data covered by the vector range are assigned a value of 1, and the remaining pixels are assigned a value of 0, to obtain different types of topographic areas at the initial resolution, and record them as the first raster data.

4. The method according to claim 3, characterized in that Quantify the marked area to generate the second raster data, and proceed as follows: Generate unvalued raster data according to a target resolution based on a range covered by the first raster data; All pixels in the unvalued raster data are traversed, and labels are assigned to each pixel to generate second raster data containing labels of different types of landforms.

5. The method according to claim 4, characterized in that The different types of topographical and geomorphic areas include: the inner area of ​​the lunar impact crater, the edge area of ​​the lunar impact crater, and other topographical and geomorphic areas; Correspondingly, all pixels in the unvalued raster data are traversed, and labels are assigned to each pixel to generate second raster data containing labels of different types of landforms, specifically: Recording the currently traversed pixel in the unvalued raster data as a first pixel, and determining a second pixel set corresponding to the first pixel according to a spatial position correspondence relationship between the unvalued raster data and the first raster data; The second pixel set is a plurality of pixels corresponding to the first pixel in the first raster data; If the values ​​of all pixels in the second pixel set are 1, the first pixel is assigned as the label of the inner area of ​​the lunar impact crater; If the values ​​of some pixels in the second pixel set are all 1, the label of the first pixel is marked as the label of the edge area of ​​the lunar impact crater; If the values ​​of all pixels in the second pixel set are not 1, the label of the first pixel is marked as the label of other topographical areas.

6. The method according to claim 1, characterized in that Based on the second raster data and the initial lunar DEM data, construct training sample data sets corresponding to different types of terrain and landforms, specifically: Dividing the second raster data into a plurality of raster data subsets according to different types of landforms; Traversing each pixel in each of the raster data subsets, extracting values ​​of multiple pixels corresponding to the currently traversed pixel from the initial lunar DEM data to form a feature vector; The feature vector is combined with the label of the currently traversed pixel to obtain a training sample data set corresponding to each of the raster data subsets.

7. The method according to claim 6, characterized in that The feature importance is multiplied with the feature vector of each training sample data set and post-processed to obtain the upscaling result of the lunar DEM at the target resolution, including: Normalizing the feature importance to obtain normalized feature importance; The normalized feature importance is multiplied by the feature vector of each training sample data set to obtain a DEM value set corresponding to different types of landforms after upscaling; The DEM value sets corresponding to the different types of landforms after upscaling are shaped to generate the upscaling results of the lunar DEM at the target resolution.

8. The method according to claim 7, characterized in that The DEM value sets corresponding to the upscaled different types of landforms are shaped to generate the upscaled lunar DEM results at the target resolution, including: Rearranging the DEM value sets corresponding to the different types of landforms after the upscaling to obtain restored raster data; According to different types of landforms, different raster data subsets are used to perform mask operations on the restored raster data, and the mask results of different types of landforms are superimposed to obtain the upscaling results of the lunar DEM at the target resolution.

9. A lunar DEM upscaling system based on expert knowledge and random forest model, characterized in that: include: a knowledge quantification unit configured to have an expert mark different types of topographic regions at an initial resolution in the initial lunar DEM data, and quantify the expert knowledge to generate second raster data; The second raster data is raster data at a target resolution, the target resolution is a resolution after upscaling, and the target resolution is lower than the initial resolution, and the second raster data includes labels of different types of landforms at the target resolution; A sample generating unit, configured to construct training sample data sets corresponding to different types of landforms based on the second raster data and the initial lunar DEM data; A model training and output unit, configured to construct a random forest model corresponding to different types of terrain and landforms according to the training sample data set, train the random forest model until convergence, and output the feature importance corresponding to the feature vector of each training sample data set; The result generating unit is configured to multiply the feature importance with the feature vector of each training sample data set correspondingly and perform post-processing to obtain the upscaling result of the lunar DEM at the target resolution.

10. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the lunar DEM upscaling method according to any one of claims 1 to 8 are implemented.

Images