A hybrid lunar surface segmentation method, system, device, and storage medium for photogrammetric point clouds in lunar rover exploration areas.

Abstract

This invention discloses a hybrid lunar surface segmentation method, system, device, and storage medium for photogrammetric point clouds in lunar rover exploration areas, belonging to the field of deep space exploration technology. The method includes: denoising and extracting initial lunar surface points from the raw scattered point cloud collected by the lunar rover in its exploration area; refining the extracted initial lunar surface points and labeling them as first lunar surface points; dividing all first lunar surface points into sub-blocks and obtaining terrain scene information for each sub-block; extracting and labeling second lunar surface points from the denoised scattered point cloud based on adaptive slope; extracting third lunar surface points from all unlabeled points based on an irregular triangular network, and labeling other points in the original scattered point cloud as non-lunar surface points; and correcting misclassified non-lunar surface points based on connected component segmentation and slope statistics, achieving accurate segmentation of lunar and non-lunar surface points. This invention can robustly filter out non-lunar surface points while preserving local lunar micro-topographical points.

Description

Technical Field

[0001] This invention belongs to the field of deep space exploration technology, specifically relating to a hybrid lunar surface segmentation method, system, device, and storage medium for photogrammetric point clouds in lunar rover exploration areas. Background Technology

[0002] As one of the closest celestial bodies to Earth in the solar system, the Moon has become a frontier site for human deep space exploration and space exploration due to its unique geographical advantages and abundant material resources. Compared with orbiter images, the binocular camera carried by the lunar rover can acquire higher resolution lunar surface images, enabling high-resolution topographic reconstruction of the lunar surface in the explored area, which is of great significance for helping humankind deepen its understanding of the lunar geological evolution history.

[0003] Currently, lunar surface topography 3D reconstruction products mainly rely on traditional digital elevation models (DEMs) and digital orthophotos. These existing 2D (or 2.5D) models lack the ability to accurately describe lunar surface features, especially the lateral features of targets, and cannot fully meet the practical needs of observation, measurement, analysis, judgment, and simulation in 3D digital space. Compared to traditional terrain products, recently developed real-world 3D models with semantic information can better represent the real physical world in a more refined manner.

[0004] However, lunar surface segmentation of photogrammetric point clouds in the rover's exploration area, a crucial foundation for constructing semantically informative 3D models of the lunar surface, has received little targeted research. Current solutions mostly employ ground-based filtering methods for existing airborne laser point clouds to achieve lunar surface segmentation. However, due to the inherent characteristics of photogrammetric point clouds and variations in the size of lunar rocks and the slope of micro-topography, existing ground-based filtering algorithms may suffer from significant misclassification when automatically extracting lunar points. Therefore, there is an urgent need to design new lunar surface segmentation methods specifically for photogrammetric point clouds in the rover's exploration area. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a hybrid lunar surface segmentation method, system, device, and storage medium for photogrammetric point clouds in lunar rover exploration areas, which can robustly filter out non-lunar surface points while preserving local lunar micro-topographic points.

[0006] This invention provides the following technical solution:

[0007] Firstly, a hybrid lunar surface segmentation method for photogrammetric point clouds in lunar rover exploration areas is provided, including the following steps:

[0008] S1: Denoise the raw scattered point cloud collected by the lunar rover in the exploration area and extract the initial lunar surface points;

[0009] S2: The extracted initial lunar points are refined based on mathematical morphology, and the refined initial lunar points are labeled as the first lunar points;

[0010] S3: Divide all first lunar surface points into sub-blocks, and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block;

[0011] S4: Combining the terrain scene information of each sub-block, extract and label the second lunar surface points from the denoised scattered point cloud based on adaptive slope;

[0012] S5: Extract the third lunar surface point from all unmarked points based on the irregular triangular network, and mark other points in the original scattered point cloud as non-lunar surface points;

[0013] S6: Based on connected component segmentation and slope statistics, misclassified non-lunar points are corrected to achieve accurate segmentation of lunar and non-lunar points.

[0014] Optionally, step S1, the specific process is as follows:

[0015] S1.1: Filtering out noise points from the original scattered point cloud using statistical analysis;

[0016] S1.2: Based on the cloth simulation algorithm, simulate the interaction between cloth nodes and the corresponding denoised scattered point cloud, determine the position of cloth surface nodes, and generate a reference lunar surface;

[0017] S1.3: Determine the distance between each point in the denoised scattered point cloud and the generated reference lunar surface. If the distance is less than the set first distance threshold, mark it as the initial lunar surface point.

[0018] Optionally, step S2, the specific process is as follows:

[0019] S2.1: Rasterize all initial lunar surface points into a two-dimensional elevation map, and filter the two-dimensional elevation map using the opening operation of mathematical morphology;

[0020] S2.2: Traverse all initial lunar surface points, determine the height difference between the initial lunar surface point and the filtered raster surface, and mark it as the first lunar surface point if it is less than the set height difference threshold.

[0021] Optionally, step S3, the specific process is as follows:

[0022] S3.1: Treat all first lunar surface points as the current input point cloud, and convert the current input point cloud into an elevation raster map by interpolation based on the average spacing of the current input point cloud;

[0023] S3.2: Calculate the elevation gradient map in the X and Y directions of the elevation raster map respectively;

[0024] S3.3: Divide the elevation gradient maps in the X and Y directions into blocks according to the set window size, and calculate the average elevation gradient of each block;

[0025] S3.4: Perform binary segmentation on the mean elevation gradient of each block, and mark the blocks with larger mean gradient values ​​as rugged terrain scenes, and mark the blocks with smaller mean gradient values ​​as flat terrain scenes.

[0026] Optionally, step S4, the specific process is as follows:

[0027] S4.1: Mark the lowest point in each block as the second lunar surface point;

[0028] S4.2: Traverse each block, taking the lowest point of each block as the center, and calculate the slope S between it and each of its surrounding neighboring points according to the following formula. adj If it is less than the terrain slope threshold S of the current block thro Then mark the neighboring point as the second lunar surface point;

[0029]

[0030] Where (X,Y,Z) are the coordinates of the lowest point of the current block, (X... adj ,Y adj Z adj ) represents the coordinates of any neighboring point of the lowest point in the current block.

[0031] Optionally, step S5, the specific process is as follows:

[0032] S5.1: Perform Delauany triangulation on all first and second lunar points to obtain the initial irregular triangular mesh of all current lunar points;

[0033] S5.2: Traverse all unmarked points and obtain the distance D between the current unmarked point n and any triangular facet δ it falls into. nδ And the angles α, β, γ between the line connecting the current unmarked point n and the three vertices of any triangular facet δ it falls into and the triangular facet δ, if the distance D nδ If the included angles α, β, and γ satisfy the following formula, then the currently unmarked point n is marked as the third lunar surface point; otherwise, it is marked as a non-lunar surface point.

[0034]

[0035] Among them, D thro and θ thro The second distance threshold and the maximum included angle threshold are set.

[0036] Optionally, step S6, the specific process is as follows:

[0037] S6.1: Divide all non-lunar points marked in step S5 into several sub-point clouds, and each sub-point cloud contains a set number of non-lunar points;

[0038] S6.2: Determine the average slope between all non-lunar points in each sub-point cloud. If the slope is less than the set threshold, then correct and mark all non-lunar points in the sub-point cloud as the third lunar point, thus achieving the segmentation of lunar points and non-lunar points.

[0039] Secondly, a hybrid lunar surface segmentation system for photogrammetric point clouds in lunar rover exploration areas is provided, including:

[0040] The initial extraction module is used to denoise and extract initial lunar surface points from the raw scattered point cloud collected by the lunar rover in the exploration area.

[0041] The first lunar point annotation module is used to refine the extracted initial lunar points based on mathematical morphology and to annotate the refined initial lunar points as the first lunar points.

[0042] The terrain information acquisition module is used to divide all the first lunar surface points into sub-blocks and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block;

[0043] The second-month surface point annotation module combines the terrain scene information of each sub-block and extracts and annotates the second-month surface points from the denoised scattered point cloud based on adaptive slope.

[0044] The third lunar facet extraction module is used to extract the third lunar facets from all unmarked points based on the irregular triangular network, and to mark other points in the original scattered point cloud as non-lunar facets.

[0045] The correction module is used to correct misclassified non-lunar points based on connected component segmentation and slope statistics, thereby achieving accurate segmentation of lunar and non-lunar points.

[0046] Thirdly, a computer device is provided, including a processor and a memory; wherein, when the processor executes a computer program stored in the memory, it implements the steps of the hybrid lunar surface segmentation method for photogrammetric point clouds of a lunar rover's exploration area as described in any one of the first aspects.

[0047] Fourthly, a computer-readable storage medium is provided for storing a computer program; when executed by a processor, the computer program implements the steps of the hybrid lunar surface segmentation method for photogrammetric point clouds of a lunar rover's exploration area as described in any one of the first aspects.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] This application denoises the original scattered point cloud and then uses cloth simulation to extract initial lunar surface points from the denoised point cloud for preliminary refinement. Next, this application further refines the extracted initial lunar surface points based on mathematical morphology. Furthermore, this application combines terrain scene information and adaptive slope to extract second lunar surface points from the denoised scattered point cloud. This operation can be flexibly adjusted according to the slope characteristics of different terrains, thereby improving the extraction accuracy of the second lunar surface points. Based on the accurate extraction of the first and second lunar surface points, an irregular triangular mesh progressive encryption technique is used to extract the third lunar surface points, which further improves the extraction accuracy. Finally, this application corrects misclassified non-lunar surface points based on connected component segmentation and slope statistics, making the final extracted lunar surface points more complete and accurate. This application employs a multi-level processing strategy from coarse to fine for hybrid segmentation of the lunar surface, which can robustly filter out non-lunar surface points while preserving local micro-topographic points on the lunar surface. Attached Figure Description

[0050] Figure 1 This is a flowchart of the hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area, as described in this invention.

[0051] Figure 2 This is an example of a photogrammetric point cloud of a lunar surface inspection area after denoising based on statistical analysis, according to the present invention.

[0052] Figure 3 This is an example of the initial lunar surface points extracted based on cloth simulation according to the present invention.

[0053] Figure 4 This is an example of the initial lunar surface point refinement based on mathematical morphology of the present invention.

[0054] Figure 5 This is an example of the non-lunar point segmentation results of the photogrammetric point cloud in the lunar surface survey area of ​​the present invention.

[0055] Figure 6 This is an example of lunar surface point segmentation results from the photogrammetric point cloud of the lunar surface survey area of ​​the present invention. Detailed Implementation

[0056] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0057] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the term "comprising" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0058] Example 1

[0059] like Figure 1 As shown, a hybrid lunar surface segmentation method for photogrammetric point clouds in a lunar rover's exploration area is provided, including the following steps:

[0060] S1: Denoise the raw scattered point cloud collected by the lunar rover in the exploration area and extract the initial lunar surface points.

[0061] Step S1 specifically includes the following sub-steps:

[0062] S1.1: Filter out noise points from the original scattered point cloud using statistical analysis.

[0063] Statistical analysis can be used for denoising, referencing existing techniques. Specifically, the average distance between each point in the original scattered point cloud and its surrounding neighbors is calculated. It is assumed that these calculated average distances generally follow a Gaussian distribution. The global average distance and standard deviation are calculated. Points whose average distance to their surrounding neighbors is greater than the sum of the global average distance and the standard deviation are marked as noise points (outliers) and removed. Figure 2 The image shown is an example of a denoised photogrammetric point cloud of a lunar surface survey area.

[0064] S1.2: Based on the cloth simulation algorithm, simulate the interaction between cloth nodes and the corresponding denoised scattered point cloud, determine the position of cloth surface nodes, and generate a reference lunar surface.

[0065] Specifically, the denoised scattered point cloud is regarded as potential cloth nodes, which are distributed in three-dimensional space and represent possible locations on the lunar surface.

[0066] According to formula (1), the movement of the cloth node in three-dimensional space is simulated;

[0067]

[0068] Where X represents the position of the particle at time t; F ext (X,t) represents the external force, which consists of the gravitational force and collision force generated by obstacles when the particle encounters certain objects in its direction of motion; F int (X,t) represents the internal force of the particle at position X and time t.

[0069] By simulating the movement of cloth nodes, the position of each node at a specific time point is obtained, and these positions together constitute an approximate representation of the cloth surface. In this process, the cloth surface gradually conforms to the obstacles in the simulated environment (the virtual lunar surface position in this application), thereby forming a surface similar in shape to the lunar surface, i.e., the reference lunar surface.

[0070] S1.3: Determine the distance between each point in the denoised scattered point cloud and the generated reference lunar surface. If the distance is less than the set first distance threshold, mark it as the initial lunar surface point.

[0071] The distance between each point and the generated reference lunar surface can be obtained using existing technology. In this application, as an option, the first distance threshold is set to 0.5m.

[0072] S2: The extracted initial lunar surface points are refined based on mathematical morphology, and the refined initial lunar surface points are labeled as the first lunar surface points.

[0073] Step S2, the specific process is as follows:

[0074] S2.1: Rasterize all initial lunar surface points into a two-dimensional elevation map, and filter the two-dimensional elevation map using the opening operation of mathematical morphology.

[0075] Opening filtering involves erosion followed by dilation.

[0076] The result of the erosion operation is to obtain the minimum elevation value within the filter window, as shown in formula (2):

[0077]

[0078] The result of the dilation operation is to obtain the maximum elevation value in the filter window, as shown in formula (3):

[0079]

[0080] The selection of the filtering window can be determined based on existing technology.

[0081] S2.2: Traverse all initial lunar surface points, determine the height difference between the initial lunar surface point and the filtered raster surface, and mark it as the first lunar surface point if it is less than the set height difference threshold.

[0082] Referring to existing technologies, the elevation difference between the initial lunar surface point and the grid surface is calculated. In this application, as an option, the elevation difference threshold is 0.1m.

[0083] S3: Divide all first lunar points into sub-blocks and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block.

[0084] Step S3, the specific process is as follows:

[0085] S3.1: Treat all first lunar surface points as the current input point cloud, and convert the current input point cloud into an elevation raster map by interpolation based on the average spacing of the current input point cloud.

[0086] The average spacing of the input point cloud, i.e. the average distance between adjacent points, is obtained using existing technology.

[0087] S3.2: Calculate the elevation gradient map in the X and Y directions of the elevation raster map respectively.

[0088] An elevation raster map contains elevation data and uses existing techniques to calculate the rate of elevation change of each pixel or grid point in the X and Y directions, visualizing the rate of elevation change as a gradient map.

[0089] S3.3: Divide the elevation gradient maps in the X and Y directions into blocks according to the set window size, and calculate the average elevation gradient of each block.

[0090] The window size is determined based on expert experience.

[0091] S3.4: Perform binary segmentation on the mean elevation gradient of each block, and mark the blocks with larger mean gradient values ​​as rugged terrain scenes, and mark the blocks with smaller mean gradient values ​​as flat terrain scenes.

[0092] That is, all blocks are marked as rugged terrain scenes or flat terrain scenes.

[0093] S4: Combining the terrain scene information of each sub-block, extract and label the second lunar surface points from the denoised scattered point cloud based on adaptive slope.

[0094] Step S4, the specific process is as follows:

[0095] S4.1: Mark the lowest point in each block as the second lunar surface point.

[0096] S4.2: Traverse each block, and using the lowest point of each block as the center, calculate the slope S between it and each surrounding neighboring point according to formula (4). adj If it is less than the terrain slope threshold S of the current block thro S adj <Sadj Then mark the neighboring point as the second lunar surface point;

[0097]

[0098] Where (X,Y,Z) are the coordinates of the lowest point of the current block, (X... adj ,Y adj Z adj () represents the coordinates of any neighboring point of the lowest point of the current block. In this application, the number of neighboring points of the lowest point of the block is set to 4.

[0099] In this application, if the terrain of the current block is flat, then S thro Set to 0.2; if the terrain of the current block is rugged, S thro Set it to 0.3.

[0100] S5: Extract the third lunar surface point from all unmarked points based on the irregular triangular network, and mark the other points in the original scattered point cloud as non-lunar surface points.

[0101] Step S5, the specific process is as follows:

[0102] S5.1: Perform Delauany triangulation on all first and second lunar points to obtain the initial irregular triangular mesh of all current lunar points.

[0103] Delauany triangulation can generate triangular meshes that maximize minimum angles, uniqueness, and local optima. Initial triangular irregular meshes (TINs) are obtained through Delauany triangulation.

[0104] S5.2: Traverse all unmarked points and obtain the distance D between the current unmarked point n and any triangular facet δ it falls into. nδ And the angles α, β, γ between the line connecting the current unmarked point n and the three vertices of any triangular facet δ it falls into and the triangular facet δ, if the distance D nδ If the included angles α, β, and γ satisfy formula (5), then the current unmarked point n is marked as the third lunar surface point; otherwise, it is marked as a non-lunar surface point.

[0105]

[0106] Among them, D thro and θ thro The second distance threshold and the maximum included angle threshold are set.

[0107] In this application, D thro It is 0.03m, θ throThe angle is 10°. Unmarked points refer to all points in the denoised scattered point cloud except for the marked first and second lunar facet points; the distance D between the current unmarked point n and any triangular facet δ it falls into. nδ The calculation method can refer to existing technology, and the method for obtaining the included angles α, β, and γ can also refer to existing technology.

[0108] S6: Based on connected component segmentation and slope statistics, misclassified non-lunar points are corrected to achieve accurate segmentation of lunar and non-lunar points.

[0109] Step S6, the specific process is as follows:

[0110] S6.1: Divide all non-lunar points marked in step S5 into several sub-point clouds, and each sub-point cloud contains a set number of non-lunar points.

[0111] In this application, each sub-point cloud contains 10 points, and the division method can refer to existing technologies.

[0112] S6.2: Determine the average slope between each non-lunar point in each sub-point cloud. If the slope is less than the set threshold, then correct and mark all non-lunar points in the sub-point cloud as the third lunar point. All non-lunar points that are not corrected and marked are the final non-lunar points, thus realizing the segmentation of lunar points and non-lunar points.

[0113] The calculation method for the average slope between each non-lunar surface point can refer to the content of formula (4). It is worth noting that the noise points removed in step S1 can be marked as noise points or non-lunar surface points as needed.

[0114] It is worth noting that there is no substantial difference between the first, second, and third lunar facets in this application. They are all lunar facets segmented at different stages. The first, second, and third are only used to distinguish similar objects. In the subsequent process of step S6, the first, second, and third lunar facets can be uniformly named to achieve the segmentation of lunar facets and non-lunar facets.

[0115] like Figures 3-6 As shown, if lunar surface points are extracted solely through cloth simulation, or if lunar surface points are extracted through cloth simulation and then refined using mathematical morphology, the robustness of the segmentation is poor compared to the hybrid lunar surface segmentation method used in this application, and it cannot preserve local micro-topographic points on the lunar surface.

[0116] Example 2

[0117] A hybrid lunar surface segmentation system for photogrammetric point clouds in lunar rover exploration areas includes:

[0118] The initial extraction module is used to filter out noise points from the original scattered point cloud collected by the lunar rover in the exploration area, and to extract the initial lunar surface points from the denoised scattered point cloud.

[0119] The first lunar point annotation module is used to refine the extracted initial lunar points based on mathematical morphology and to annotate the refined initial lunar points as the first lunar points.

[0120] The terrain information acquisition module is used to divide all the first lunar surface points into sub-blocks and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block;

[0121] The second-month surface point annotation module combines the terrain scene information of each sub-block and extracts and annotates the second-month surface points from the denoised scattered point cloud based on adaptive slope.

[0122] The third lunar facet extraction module is used to extract the third lunar facets from all unmarked points based on the irregular triangular network, and to mark other points in the original scattered point cloud as non-lunar facets.

[0123] The correction module is used to correct misclassified non-lunar points based on connected component segmentation and slope statistics, thereby achieving accurate segmentation of lunar and non-lunar points.

[0124] For more detailed information on the above methods, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0125] Example 3

[0126] The present invention provides a computer device, including a processor and a memory; wherein, when the processor executes the computer program stored in the memory, it implements the steps of the above-described hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area.

[0127] For more detailed information on the above methods, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0128] Example 4

[0129] The present invention provides a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, it implements the steps of the above-described hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area.

[0130] For more detailed information on the above methods, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0131] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. The systems, devices, and storage media disclosed in the embodiments are described simply because they correspond to the methods disclosed in the embodiments; relevant details can be found in the method section.

[0132] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0133] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A hybrid lunar surface segmentation method for photogrammetric point clouds in a lunar rover's exploration area, characterized in that, Includes the following steps: S1: Denoise the raw scattered point cloud collected by the lunar rover in the exploration area and extract the initial lunar surface points; S2: The extracted initial lunar points are refined based on mathematical morphology, and the refined initial lunar points are labeled as the first lunar points; S3: Divide all first lunar surface points into sub-blocks, and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block; S4: Combining the terrain scene information of each sub-block, extract and label the second lunar surface points from the denoised scattered point cloud based on adaptive slope; S5: Extract the third lunar surface point from all unmarked points based on the irregular triangular network, and mark other points in the original scattered point cloud as non-lunar surface points; S6: Based on connected component segmentation and slope statistics, misclassified non-lunar points are corrected to achieve accurate segmentation of lunar and non-lunar points.

2. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S1, the specific process is as follows: S1.1: Filtering out noise points from the original scattered point cloud using statistical analysis; S1.2: Based on the cloth simulation algorithm, simulate the interaction between cloth nodes and the corresponding denoised scattered point cloud, determine the position of cloth surface nodes, and generate a reference lunar surface; S1.3: Determine the distance between each point in the denoised scattered point cloud and the generated reference lunar surface. If the distance is less than the set first distance threshold, mark it as the initial lunar surface point.

3. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S2, the specific process is as follows: S2.1: Rasterize all initial lunar surface points into a two-dimensional elevation map, and filter the two-dimensional elevation map using the opening operation of mathematical morphology; S2.2: Traverse all initial lunar surface points, determine the height difference between the initial lunar surface point and the filtered raster surface, and mark it as the first lunar surface point if it is less than the set height difference threshold.

4. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S3, the specific process is as follows: S3.1: Treat all first lunar surface points as the current input point cloud, and convert the current input point cloud into an elevation raster map by interpolation based on the average spacing of the current input point cloud; S3.2: Calculate the elevation gradient map in the X and Y directions of the elevation raster map respectively; S3.3: Divide the elevation gradient maps in the X and Y directions into blocks according to the set window size, and calculate the average elevation gradient of each block; S3.4: Perform binary segmentation on the mean elevation gradient of each block, and mark the blocks with larger mean gradient values ​​as rugged terrain scenes, and mark the blocks with smaller mean gradient values ​​as flat terrain scenes.

5. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S4, the specific process is as follows: S4.1: Mark the lowest point in each block as the second lunar surface point; S4.2: Traverse each block, taking the lowest point of each block as the center, and calculate the slope S between it and each of its surrounding neighboring points according to the following formula. adj If it is less than the terrain slope threshold S of the current block thro Then mark the neighboring point as the second lunar surface point; Where (X,Y,Z) are the coordinates of the lowest point of the current block, (X... adj ,Y adj Z adj ) represents the coordinates of any neighboring point of the lowest point in the current block.

6. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S5, the specific process is as follows: S5.1: Perform Delauany triangulation on all first and second lunar points to obtain the initial irregular triangular mesh of all current lunar points; S5.2: Traverse all unmarked points and obtain the distance D between the current unmarked point n and any triangular facet δ it falls into. nδ And the angles α, β, γ between the line connecting the current unmarked point n and the three vertices of any triangular facet δ it falls into and the triangular facet δ, if the distance D nδ If the included angles α, β, and γ satisfy the following formula, then the currently unmarked point n is marked as the third lunar surface point; otherwise, it is marked as a non-lunar surface point. Among them, D thro and θ thro The second distance threshold and the maximum included angle threshold are set.

7. The hybrid lunar surface segmentation method for photogrammetric point clouds in the lunar rover's exploration area according to claim 1, characterized in that, Step S6, the specific process is as follows: S6.1: Divide all non-lunar points marked in step S5 into several sub-point clouds, and each sub-point cloud contains a set number of non-lunar points; S6.2: Determine the average slope between all non-lunar points in each sub-point cloud. If the slope is less than the set threshold, then correct and mark all non-lunar points in the sub-point cloud as the third lunar point, thus achieving the segmentation of lunar points and non-lunar points.

8. A hybrid lunar surface segmentation system for photogrammetric point clouds in a lunar rover's exploration area, characterized in that, include: The initial extraction module is used to denoise and extract initial lunar surface points from the raw scattered point cloud collected by the lunar rover in the exploration area. The first lunar point annotation module is used to refine the extracted initial lunar points based on mathematical morphology and to annotate the refined initial lunar points as the first lunar points. The terrain information acquisition module is used to divide all the first lunar surface points into sub-blocks and obtain the terrain scene information of each sub-block based on the elevation difference statistics within each sub-block; The second-month surface point annotation module combines the terrain scene information of each sub-block and extracts and annotates the second-month surface points from the denoised scattered point cloud based on adaptive slope. The third lunar facet extraction module is used to extract the third lunar facets from all unmarked points based on the irregular triangular network, and to mark other points in the original scattered point cloud as non-lunar facets. The correction module is used to correct misclassified non-lunar points based on connected component segmentation and slope statistics, thereby achieving accurate segmentation of lunar and non-lunar points.

9. A computer device, characterized in that, It includes a processor and a memory; wherein, when the processor executes the computer program stored in the memory, it implements the steps of the hybrid lunar surface segmentation method for photogrammetric point clouds of the lunar rover's exploration area as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, Used to store computer programs; when the computer programs are executed by a processor, they implement the steps of the hybrid lunar surface segmentation method for photogrammetric point clouds of the lunar rover's exploration area as described in any one of claims 1-7.

Images