A method and device for multi-resolution digital elevation model 3D visualization

Abstract

The application discloses a kind of multi-resolution digital elevation model 3D visualization method and device, comprising: one, data preprocessing stage: S1, obtain digital elevation model DEM data;S2, calculate the multiple terrain factors of digital elevation model DEM data, principal component analysis, and the terrain feature comprehensive index is calculated;Two, real-time data processing stage: S3, multiple evaluation factor quadtree construction;S4, multiple evaluation factor quadtree updates;S5, terrain visualization: based on the multiple evaluation factor quadtree, screen space clipping is carried out, occlusion is removed, feature evaluation is carried out, obtains all points needing to be loaded, constructs irregular triangle net, is sent to GPU rendering, loads picture.

Description

Technical Field

[0001] This invention relates to the field of computer graphics technology, specifically to a method and apparatus for three-dimensional visualization of multi-resolution digital elevation models. Background Technology

[0002] A Digital Elevation Model (DEM) is a digital representation of Earth's surface geographic information in a two-dimensional geographic space using a Geographic Information System (GIS). It typically uses a two-dimensional array to represent spatial information and is commonly used for terrain analysis. However, understanding its use in terrain visualization and analysis requires a certain level of specialized knowledge and is less intuitive and efficient than three-dimensional models. To better apply DEMs to terrain analysis, flight simulation, and military simulation, current research focuses on how to perform three-dimensional spatial visualization and reconstruction of two-dimensional spatial data acquired by satellites or other means.

[0003] 3D terrain visualization comprises two parts: terrain simplification and visibility computation. The mainstream approach to terrain simplification is based on quadtrees to construct Level of Detail (LOD) models, allowing for the display of models at different resolutions depending on various conditions. Visibility computation reduces the rendering of invisible data at the current location by using information such as camera position and occlusion relationships, improving the real-time performance of model visualization. Research primarily focuses on occlusion culling. Current research faces two main problems: First, quadtree algorithms typically select only the four corner points or the midpoints of the edges of a rectangular region as quadtree nodes. While this simplifies the terrain, these points are often not terrain feature points, resulting in limited accuracy. To achieve better model accuracy, the quadtree is repeatedly divided, leading to a continuous increase in tree height. The constructed quadtree is often not a balanced quadtree, and the different resolutions of adjacent plots cause terrain cracks during final mesh construction. Furthermore, the time overhead of traversing and updating the quadtree is significant. Second, occlusion culling in terrain visualization is only suitable for image space-based methods, such as the Z-Buffer algorithm and ray casting algorithm. Although these offer high accuracy, they are computationally intensive and prone to redundant calculations.

[0004] Solutions for terrain simplification typically employ filtering or terrain feature-based methods to obtain feature points, then use irregular grids to divide the terrain or traverse quadtrees multiple times to determine the resolution of adjacent regions, patching low-resolution areas. However, filtering methods can lead to overly smooth terrain effects that lose realism. Terrain feature methods, which use a few terrain feature indicators as evaluation metrics, can result in a lack of smoothness. Some studies have also used spatial autocorrelation methods, but these are computationally complex and computationally intensive.

[0005] The solution for visibility computation accelerates the computation by adding bounding boxes to the target of light projection. Common bounding boxes include spherical bounding boxes and cuboid bounding boxes. However, in terrain visualization, it does not take into account that mountains are often cone-shaped and have continuous spatial characteristics, which can lead to problems such as excessive occlusion and low utilization of bounding box space. Summary of the Invention

[0006] To address the above issues, this invention provides a multi-resolution digital elevation model (DEM) 3D visualization method. It innovatively proposes a multi-evaluation factor quadtree hierarchical detail model based on terrain blocks. Principal Component Analysis (PCA) is performed on commonly used terrain factors to construct a comprehensive terrain feature index, quantifying the intensity of terrain features. Feature point selection is determined based on multiple evaluation criteria and occlusion relationships. Finally, a TIN network is constructed to achieve crack-free 3D terrain visualization.

[0007] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0008] The technical solution of this invention is divided into two parts: preprocessing and real-time processing. The preprocessing part includes the calculation of the comprehensive index of terrain features, and the real-time processing part includes the initialization and updating of the multi-evaluation factor quadtree, screen space pruning, occlusion removal, and feature point selection.

[0009] To avoid the space overhead of recursive quadtree construction, a queue is used to store the entire quadtree. A level pointer array is used to store the starting address of each node at each level, accelerating the search for leaf nodes. A subdivision marker array is used to store the indices of subdivided nodes at each level, facilitating quadtree updates. Each node stores only three pieces of information: the center coordinates of the terrain tile it represents, a marker indicating whether the node has been subdivided, and the size of the terrain tile it represents.

[0010] First, input the digital elevation model (DEM) file, calculate multiple terrain factors of the original DEM, perform principal component analysis on these factors, and construct a comprehensive terrain feature index as input for the next step.

[0011] Then, a node evaluation system is constructed, divided into a static part and a dynamic part. The static part includes tree height evaluation criteria and land use evaluation criteria, which serve as the basic evaluation criteria for the multi-evaluation factor quadtree and do not require updating; the dynamic part includes line-of-sight evaluation criteria, field-of-view evaluation criteria, and feature evaluation criteria, which are updated in real time along with camera parameters.

[0012] The tree height evaluation criterion is determined by the width of the original terrain and the model simplification rate, and is used to construct a simple quadtree. The land type evaluation criterion is determined by three types of terrain: plains, hills, and mountains. The terrain type of the current area is determined by calculating the height of multiple nodes and child nodes, and a land type quadtree is constructed based on the simple quadtree. The view distance evaluation criterion is determined by the farthest and nearest effective distances, and is the basis of the dynamic part of the node evaluation system. The field of view evaluation criterion is determined by the range to be observed, resulting in a viewpoint quadtree containing the observation center area, transition area, and non-observation center area, completing the initialization of the multi-evaluation factor quadtree. The feature evaluation criterion changes in real time according to the view distance evaluation criterion, and feature points are dynamically acquired. To reduce the amount of computation, feature point acquisition needs to be performed after screen space clipping and occlusion culling.

[0013] After initializing the multi-evaluation factor quadtree, terrain visualization is required, including screen-space clipping and occlusion culling. During camera navigation, much terrain is off-screen and cannot be seen. Ray casting to all terrain tiles would result in a large amount of unnecessary computation; therefore, screen-space clipping is necessary for off-screen points. First, objects are transformed from world space coordinates to standard device coordinates. Then, the x and y values ​​of the standard device coordinates are used to determine if the object appears on the screen. Further processing is only performed on terrain tiles that do appear on the screen.

[0014] Occlusion culling is based on the following premise: When a tall mountain appears on the screen, and the camera is far from the ground, the number of pixels the mountain occupies on the screen decreases. The viewing distance evaluation criterion and feature evaluation criterion also reduce the number of feature points for each terrain tile. Conversely, when the camera is close to the ground, the number of terrain tiles appearing on the screen after screen space clipping is also limited. Therefore, as long as the camera's line of sight forms a 90-degree angle with the horizontal plane in the first frame, it can be ensured that there is almost no complete occlusion of terrain tiles on the screen at this time, and occlusion culling is unnecessary. An occlusion triangle is constructed for each terrain tile represented by a leaf node and stored in the occlusion array, ready for occlusion relationship judgment in the next frame. When camera parameters change, a ray is emitted from the center of the terrain tile to the camera position. The position of this ray on the screen is calculated, and the corresponding occlusion array is read. The occlusion array is iterated to see if there is an occlusion triangle intersecting the ray. If there is, it means the terrain tile is occluded, and no new feature points need to be added. If not, it means the terrain tile is not occluded, and an occlusion triangle for the next frame is added to the occlusion array.

[0015] The leaf nodes after occlusion culling represent a rough terrain block model. Feature points need to be added according to the feature evaluation criteria to improve the accuracy of the terrain model without increasing the tree height. Finally, all the vertices of the 3D model that need to be rendered are used to construct a TIN and send it to the GPU for rendering.

[0016] Firstly, a method for 3D visualization of multi-resolution digital elevation models is provided, including:

[0017] I. Data Preprocessing Stage:

[0018] S1. Obtain Digital Elevation Model (DEM) data;

[0019] S2. Calculate multiple terrain factors from the digital elevation model (DEM) data, perform principal component analysis (PCA) on these terrain factors, and calculate the comprehensive terrain feature index.

[0020] II. Real-time Data Processing Stage:

[0021] S3. Construction of a quadtree with multiple evaluation factors: Based on the digital elevation model (DEM) data and the comprehensive indicators of terrain features, the tree height evaluation criteria, land use evaluation criteria, sight distance evaluation criteria, field of view evaluation criteria and feature evaluation criteria are constructed in sequence to obtain a quadtree with multiple evaluation factors.

[0022] S4. Multi-evaluation factor quadtree update: As camera parameters change during roaming, the line-of-sight evaluation criteria, field-of-view evaluation criteria, and feature evaluation criteria are updated in real time. Line-of-sight and field-of-sight evaluations are then performed again to obtain a new multi-evaluation factor quadtree.

[0023] S5. Terrain Visualization: Based on the multi-evaluation factor quadtree, screen space pruning, occlusion removal, and feature evaluation are performed to obtain all points that need to be loaded. An irregular triangular network (TIN) is constructed and sent to the GPU for rendering to load the screen.

[0024] In some embodiments, the topographic factors include slope, surface roughness, surface cutting depth, topographic relief, elevation variation coefficient, profile curvature, and plane curvature.

[0025] In some embodiments, constructing tree height evaluation criteria includes:

[0026] Based on the DEM size and the comprehensive ratio of terrain feature indicators of terrain feature points, a tree height evaluation criterion Level is constructed:

[0027] Level = log2 W - log2 Rh

[0028] Where W represents the DEM size, R represents the maximum proportion of points selected by the comprehensive index of terrain features when the terrain error is less than a certain threshold, and h represents the parameter that can be adjusted autonomously; a quadtree that meets the tree height evaluation criteria is called a simple quadtree.

[0029] Constructing land use evaluation criteria, including:

[0030] Calculate the height of multiple nodes and child nodes to determine the terrain category of the current area, and construct a terrain category quadtree based on a simple quadtree;

[0031] Take the center point of the node and the center points of its four child nodes after partitioning the node once.

[0032] If the height of any one of these 5 points is greater than or equal to 200 meters and less than 500 meters, then the area is considered to be hilly terrain and is divided into sections.

[0033] If the altitude is greater than or equal to 500 meters, the area is considered mountainous and is divided twice.

[0034] If the distance is less than 200 meters, the area is considered a plain and does not need to be divided. The final result shows that the three types of terrain differ by only 1 level in the quadtree, which facilitates subsequent network construction and avoids cracks.

[0035] In some embodiments, a line-of-sight evaluation criterion is constructed, including:

[0036] First, the maximum effective distance Far is the distance at which the terrain is exactly displayed on the screen:

[0037]

[0038] Where W represents the side length of the DEM and Fov represents the camera parameters;

[0039] Then, the nearest effective distance is defined as the distance at which a small portion of the terrain is exactly displayed on the screen:

[0040] Near = Far × α

[0041] Where α represents the scaling factor;

[0042] Finally, based on the farthest effective distance (Far) and the nearest effective distance (Near), the line-of-sight evaluation criterion Distance is constructed:

[0043]

[0044] Where Camera represents the current distance between the camera and the center of observation;

[0045] And / or, construct vision evaluation criteria, including:

[0046] The observation area is divided into a central observation area (Inner) and a transition area (Out).

[0047]

[0048] Inner = Out × 0.5

[0049] Where Outmax Out represents the maximum side length of the outer transition region. min Far represents the minimum side length of the outer transition zone, Near represents the farthest effective distance, Distance represents the line-of-sight evaluation criterion.

[0050] Construct feature evaluation criteria, including:

[0051] Based on the comprehensive terrain feature index and the sight distance evaluation criterion, the feature evaluation criterion Threshold is constructed as follows:

[0052]

[0053] Where F min F represents the minimum comprehensive index of terrain features. max The maximum terrain feature is represented by the comprehensive index, Distance represents the line-of-sight evaluation criterion, Far represents the farthest effective distance, and Near represents the nearest effective distance.

[0054] In some embodiments, initialization to obtain a multi-evaluation factor quadtree includes:

[0055] Based on the digital elevation model (DEM) data, the entire terrain area is used as the root node of the quadtree. It is determined whether the quadtree meets the tree height evaluation criteria. If it does, it is further subdivided; otherwise, it is stopped, resulting in a simple quadtree representing the basic terrain.

[0056] Then, based on the land use evaluation criteria, the leaf nodes of the simple quadtree were further subdivided to obtain a land use quadtree;

[0057] Finally, determine whether the leaf nodes of the land type quadtree meet the view evaluation criteria: if they do, it means that the node is in the observation area and needs to be further divided; if they do not meet the criteria, no division is needed, and the viewpoint quadtree is obtained. The construction of the multi-evaluation factor quadtree is completed.

[0058] In some embodiments, updating the multi-evaluation factor quadtree includes:

[0059] After camera parameters change, first update the line-of-sight evaluation criterion, then update the field-of-view evaluation criterion. Based on the new field-of-sight evaluation criterion, determine whether all nodes need to be subdivided or deleted; this is divided into the following four cases:

[0060] 1) Nodes that previously met the field of view evaluation criteria do not need to be updated if they still meet the criteria after camera parameter changes;

[0061] 2) If a leaf node that previously met the field of view evaluation criteria no longer meets the criteria after the camera parameters change, it needs to be deleted. At the same time, the parent node of the leaf node is remarked as unsubdivided and becomes a new leaf node.

[0062] 3) Nodes that previously did not meet the field of view evaluation criteria do not need to be updated if they still do not meet the criteria after camera parameter changes;

[0063] 4) Nodes that previously did not meet the field of view evaluation criteria, but now meet the criteria after camera parameter changes, need to be subdivided. At the same time, the node is marked as subdivided and four leaf nodes are added.

[0064] In some embodiments, screen space cropping includes:

[0065] Determine whether a leaf node of the multi-evaluation factor quadtree is within the screen. If it is, proceed to the next step; otherwise, skip the node.

[0066] In some embodiments, occlusion culling includes:

[0067] Based on the leaf nodes on the screen, a ray is emitted towards the camera position. The position of the ray on the screen is calculated, and the occlusion array at the corresponding position is read. It is determined whether there is an occlusion triangle intersecting with the ray. If there is, it means that the terrain tile is occluded and no new feature point needs to be added. If not, it means that the terrain tile is not occluded. The height of the occlusion triangle is taken as the height of the center point of the terrain tile, and half the size of the terrain tile is taken as the length of the base of the occlusion triangle. The occlusion array is then added.

[0068] The updated quadtree needs to undergo occlusion culling again, and the steps are as follows:

[0069] 1. Project the node's position in the world coordinate system onto the screen space, divide the space, and determine its position in the screen space;

[0070] 2. In the world space coordinate system, take the location of this node as the source point, and the line connecting the source point and the camera position as the direction ray;

[0071] 3. Based on the occlusion triangle set obtained during initialization, read the occlusion array in the screen space where the node is located;

[0072] 4. Traverse the occlusion triangles in the array, and check the intersection of the ray and the occlusion triangle. If they intersect, it means that the node is occluded and no rendering is needed. Then check the next node.

[0073] 5. If the node is not occluded, construct the occlusion triangle for that node and store it in a temporary array. After all nodes have been judged, use the temporary array as a new set of occlusion triangles for the next frame.

[0074] In some embodiments, feature evaluation includes:

[0075] Based on the unoccluded leaf nodes, determine whether the points within the range represented by the node meet the feature evaluation criteria. If they do, add them to the rendering queue; otherwise, skip them. Finally, obtain all the required points, construct a TIN, and send it to the GPU for rendering.

[0076] In a second aspect, the present invention provides a three-dimensional visualization device for a multi-resolution digital elevation model, including a processor and a storage medium;

[0077] The storage medium is used to store instructions;

[0078] The processor is configured to operate according to the instructions to perform the steps of the method according to the first aspect.

[0079] Thirdly, the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0080] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0081] (1) In this invention, PCA dimensionality reduction is performed on seven commonly used terrain factors to obtain a comprehensive terrain feature index, which serves as a quantification of terrain feature intensity. The comprehensive terrain feature index is a scalar; the larger the absolute value, the more obvious the terrain feature. Specifically, a larger value indicates a more obvious foothill feature at that point, while a smaller value indicates a more obvious ridge or valley feature. Each point in the DEM has a corresponding comprehensive terrain feature index value. Compared to single feature points (such as ridge points or valley points) or filtering methods (such as the importance point method), points obtained through the comprehensive terrain feature index have obvious continuous features, can comprehensively describe terrain features, and better preserve the topography.

[0082] (2) The present invention selects multiple regions and obtains the average RMSE of the following different methods: 1.304 for ridge point alone, 1.294 for valley point alone, 1.182 for ridge and valley point combined, and 2.041 for Very Important Point (VIP). The present invention achieves 0.815 when only 70% of the feature points are loaded, indicating that the accuracy of the present invention is higher than that of the ridge and valley point method and the VIP method.

[0083] (3) The multi-evaluation factor quadtree in this invention can greatly improve rendering performance. Comparing the time overhead of the following methods in the same terrain area: real-time continuous LOD loading time is 861.009ms, average update time is 20.243ms; traditional multi-resolution loading time is 243.36ms, average update time is 16.518ms; the loading time of this invention is 23.011ms, average update time is 9.391ms. Comparing the roaming frame rate of the following methods in the same terrain area: real-time continuous LOD is 25-104 FPS, average 41 FPS; multi-resolution LOD is 26-113 FPS, average 54 FPS; this invention is 49-144 FPS, average 111 FPS. This demonstrates that this invention significantly reduces LOD construction time while also providing more stable overall performance.

[0084] (4) The conservative occlusion culling algorithm of this invention, by constructing occlusion triangles, can ensure that terrain that is not completely occluded can be displayed at a lower resolution, thus reducing the time overhead of simple light projection. After enabling occlusion culling, the test frame rate of this invention is 76-144 FPS, and the average data simplification rate reaches 31.40%, which can effectively reduce the amount of data. Attached Figure Description

[0085] Figure 1 A flowchart of a global processing method provided in an embodiment of the present invention.

[0086] Figure 2 This is a flowchart of the preprocessing process in an embodiment of the present invention.

[0087] Figure 3 This is a flowchart illustrating the construction process of a quadtree with multiple evaluation factors in an embodiment of the present invention.

[0088] Figure 4 This is a flowchart of the multi-evaluation factor quadtree update process in an embodiment of the present invention.

[0089] Figure 5 This is a flowchart illustrating the terrain visualization process in an embodiment of the present invention. Detailed Implementation

[0090] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0091] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0092] In the description of this invention, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0093] Example 1

[0094] A method for 3D visualization of multi-resolution digital elevation models includes:

[0095] I. Data Preprocessing Stage:

[0096] S1. Obtain Digital Elevation Model (DEM) data;

[0097] S2. Calculate multiple terrain factors from the digital elevation model (DEM) data, perform principal component analysis (PCA) on these terrain factors, and calculate the comprehensive terrain feature index.

[0098] II. Real-time Data Processing Stage:

[0099] S3. Construction of a quadtree with multiple evaluation factors: Based on the digital elevation model (DEM) data and the comprehensive indicators of terrain features, the tree height evaluation criteria, land use evaluation criteria, sight distance evaluation criteria, field of view evaluation criteria and feature evaluation criteria are constructed in sequence. The DEM data is then evaluated for tree height, land use, sight distance and field of view in sequence to obtain a quadtree with multiple evaluation factors.

[0100] S4. Multi-evaluation factor quadtree update: As camera parameters change during roaming, the line-of-sight evaluation criteria, field-of-view evaluation criteria, and feature evaluation criteria are updated in real time. The line-of-sight evaluation and field-of-sight evaluation are re-performed to obtain a new multi-evaluation factor quadtree.

[0101] S5. Terrain Visualization: Based on the multi-evaluation factor quadtree, screen space pruning, occlusion removal, and feature evaluation are performed to obtain all points that need to be loaded. An irregular triangular network (TIN) is constructed and sent to the GPU for rendering to load the screen.

[0102] In this embodiment, a method for 3D visualization of a multi-resolution digital elevation model (DEM) is provided. The DEM data comes from publicly available data from a geospatial data cloud, with an image size of 3601*3601 pixels and a resolution of 30m. The rendering engine used is Three.js, and the global processing flow is as follows: Figure 1 As shown.

[0103] Data preprocessing requires calculating comprehensive indicators of terrain features, such as... Figure 2 As shown.

[0104] First, seven indicators—slope, surface roughness, surface cutting depth, topographic relief, elevation variation coefficient, profile curvature, and plane curvature—were calculated from the acquired raw DEM data. Each pixel was used as a sample, and the seven indicators were used as variables to construct a two-dimensional array. Plains, hills, and mountains, each with a pixel size of 100*100 pixels, were selected as sampling areas. Each type of area was randomly selected twice to ensure data reliability. After normalizing the sampled data, KMO and Bartlett's sphericity tests were performed. The results are shown in Table 1. The KMO values ​​for all data were above 0.8, indicating strong correlation and allowing for PCA dimensionality reduction.

[0105] Table 1. Results of principal component suitability test in the sampled areas

[0106] area Plain 1 Plains 2 Hills 1 Hills 2 Mountain 1 Mountain 2 KMO 0.810 0.818 0.873 0.869 0.868 0.825

[0107] The subsequent steps are illustrated using the dimensionality reduction results of Plain 1, and the same applies to other regions. The interpretation of total variance is shown in Table 2. With two eigenvalues, the cumulative contribution rate of variance can reach 85%, indicating that two principal components are sufficient to extract the main information components. Therefore, these two principal components, F1 and F2, are selected as new indicators.

[0108] Table 2 shows the principal component suitability test results for the sampled plain 1 region provided in the embodiments of this application.

[0109] Table 2. Explanation of Total Variance

[0110]

[0111] F1 and F2 can be obtained by linear combination of the original variables, and the coefficients are the component matrix shown in Table 3. The final calculation method of the comprehensive topographic feature index is shown in Equation (1).

[0112] Table 3 is the component matrix of the sampling plain 1 region provided in the embodiments of this application.

[0113] Table 3. Component Matrix

[0114]

[0115]

[0116] This application uses a comprehensive terrain feature index F. 综合 The calculation method is as follows:

[0117]

[0118] Where F i (i = 1, 2, ..., k) represents the principal components, λ i (i = 1, 2, ..., k) are the eigenvalues, k is the number of principal components selected, and p is the total number of principal components.

[0119] Then, input the DEM file and the comprehensive terrain feature index, and construct the tree height evaluation criterion Level according to the DEM size and simplification rate, as shown in Equation (2).

[0120] Level = log2 W - log2 Rh (2)

[0121] Where W represents the DEM size, R represents the maximum proportion of points selected by the comprehensive index of terrain features when the terrain error is less than a certain threshold, and h represents the parameter that can be adjusted in height.

[0122] In this embodiment, the error threshold is set to 10m. The higher the accuracy requirement, the more feature points need to be selected, and the greater the computational cost. By adjusting the parameter h, which represents the height, the tree height can be adjusted according to the required resolution.

[0123] Then, a land use evaluation criterion is constructed. The center point of a node and the center points of the four child nodes after the node is divided once are selected. If the height of any of these five points is greater than or equal to 200 meters and less than 500 meters, the area is considered to be hilly terrain and is divided once. If it is greater than or equal to 500 meters, the area is considered to be mountainous terrain and is divided twice. If it is less than 200 meters, the area is considered to be plain and no division is required. Finally, the three types of terrain differ by only 1 level in the quadtree, which facilitates subsequent network construction and avoids gaps.

[0124] Then, a line-of-sight evaluation criterion is constructed, with the distance when the terrain is exactly fully displayed on the screen as the farthest effective distance Far, calculated as shown in Equation (3);

[0125] This application provides a maximum effective distance in its embodiments:

[0126]

[0127] Where W represents the side length of the DEM and Fov represents the camera parameters.

[0128] The nearest effective distance is taken as the distance when a small area of ​​the terrain is exactly displayed on the screen. The calculation method is shown in equation (4).

[0129] This application provides a nearest effective distance (Near):

[0130] Near = Far × α (4)

[0131] Where α represents the scaling factor.

[0132] In this embodiment, α is set to 1 / 20, which means that the closest distance between the camera position and the ground is when 1 / 20 of the terrain side length is exactly displayed on the screen.

[0133] Then, the distance evaluation criterion Distance is constructed according to equation (5):

[0134] This application provides a line-of-sight evaluation criterion, Distance:

[0135]

[0136] Where Camera represents the distance between the camera and the center point of observation.

[0137] Based on the previous line-of-sight evaluation criteria, a field-of-view evaluation criterion is constructed, dividing the observation area into a central observation area and a transition area. The central observation area is the area with the highest resolution, while the transition area serves as a transition between the central area and other low-resolution areas, avoiding abrupt changes caused by excessive resolution differences.

[0138] This application provides a visual field evaluation criterion: Out.

[0139]

[0140] Where Out max Out represents the maximum side length of the outer transition region. min Far represents the minimum side length of the outer transition zone, Far represents the farthest effective distance of equation (3), Near represents the nearest effective distance of equation (4), and Distance represents the line-of-sight evaluation criterion of equation (4).

[0141] Therefore, in this embodiment, 1 / 4 of the total terrain side length is used as the maximum side length of the outer transition zone, i.e., Out. max The parameter uses 1 / 10 of the terrain as the minimum side length of the outer transition zone, i.e., Out. min The parameters are calculated according to formula (6) to determine the side length of the transition region. In this embodiment, the side length of the inner center observation region is set to half of the outer transition region.

[0142] Then, construct the feature evaluation criterion Threshold:

[0143]

[0144] Based on the current location, determine which feature points should be added to the terrain, as shown in Equation (7). The larger the comprehensive terrain feature index, the fewer feature points need to be added; the smaller the comprehensive terrain feature index, the more feature points need to be added. Where F... min The minimum comprehensive index of terrain features is determined by the proportion of feature points represented by R in equation (3), F max The maximum terrain feature comprehensive index is represented by the comprehensive terrain feature index when 20% of the feature points are loaded in this embodiment. Distance represents the current position, and Far and Near represent the results of equations (3) and (4) respectively.

[0145] The quadtree is initialized according to the evaluation criteria, with the entire terrain area as the root node. It is then determined whether the quadtree meets the tree height evaluation criterion. If it does, further subdivision is performed; otherwise, the process stops, resulting in a simple quadtree representing the basic terrain. Next, the leaf nodes of the simple quadtree are further subdivided according to the land use evaluation criterion, resulting in a land use quadtree. Finally, it is determined whether the leaf nodes of the land use quadtree meet the view evaluation criterion: if they do, the node is within the observation area and further subdivision is needed; otherwise, no further subdivision is required, resulting in a viewpoint quadtree. The multi-evaluation factor quadtree construction is complete. The flowchart is shown below. Figure 3 As shown.

[0146] After initialization, the multi-evaluation factor quadtree first performs screen space culling, which calculates whether the current quadtree leaf node is within the screen. If it is, occlusion is culled; otherwise, the node is skipped. Since the camera's viewpoint is vertically downward and close to the ground during initialization, the terrain represented by the currently screen-space-culled node can be considered unoccluded. Due to spatial and temporal continuity, the occlusion triangle in the next frame is often the unoccluded part of the previous frame. Therefore, an occlusion triangle can be constructed for the current leaf node and stored in the occlusion array. In this embodiment, half the height of the center point of each terrain tile is used as the height of the occlusion triangle, and half the size of the terrain tile is used as the length and width of the occlusion triangle to construct a cube, which is then added to the occlusion array as the occlusion triangle.

[0147] Finally, based on the feature evaluation criteria, a certain number of feature points are added to construct a TIN, which is then sent to the GPU for rendering. After initialization, the first frame is loaded.

[0148] Multi-evaluation factor quadtree update, such as Figure 4As shown, after the camera parameters change, the viewing distance evaluation criterion is updated first, followed by the field of view evaluation criterion. Based on the new field of view evaluation criterion, it is determined whether all nodes need to be subdivided or deleted. Nodes that need to be deleted have their parent nodes become new leaf nodes after deletion, while nodes that need to be subdivided are divided into four new leaf nodes. Then, screen space culling and occlusion removal are performed again. Finally, feature points are added after feature evaluation, a new TIN is constructed, and it is sent to the GPU for rendering to form a new 3D model, as shown. Figure 5 As shown.

[0149] Example 2

[0150] Secondly, this embodiment provides a three-dimensional visualization device for a multi-resolution digital elevation model, including a processor and a storage medium;

[0151] The storage medium is used to store instructions;

[0152] The processor is configured to operate according to the instructions to perform the steps of the method according to Embodiment 1.

[0153] Example 3

[0154] Thirdly, this embodiment provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0155] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0156] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0157] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0159] As is known from common technical knowledge, this invention can be implemented through other embodiments that do not depart from its spirit or essential characteristics. Therefore, the disclosed embodiments described above are merely illustrative in all respects and are not the only ones. All modifications within the scope of this invention or its equivalents are included in this invention.

Claims

1. A method for three-dimensional visualization of a multi-resolution digital elevation model, characterized in that, include: I. Data Preprocessing Stage: S1. Obtain Digital Elevation Model (DEM) data; S2. Calculate multiple terrain factors from the digital elevation model (DEM) data, perform principal component analysis on these terrain factors, and calculate the comprehensive terrain feature index. II. Real-time Data Processing Stage: S3. Construction of a quadtree with multiple evaluation factors: Based on the digital elevation model (DEM) data and the comprehensive indicators of terrain features, the tree height evaluation criteria, land use evaluation criteria, sight distance evaluation criteria, field of view evaluation criteria and feature evaluation criteria are constructed in sequence. The DEM data is then evaluated for tree height, land use, sight distance and field of view in sequence to obtain a quadtree with multiple evaluation factors. S4. Multi-evaluation factor quadtree update: As camera parameters change during roaming, the line-of-sight evaluation criteria, field-of-view evaluation criteria, and feature evaluation criteria are updated in real time. The line-of-sight evaluation and field-of-sight evaluation are re-performed to obtain a new multi-evaluation factor quadtree. S5. Terrain Visualization: Based on the multi-evaluation factor quadtree, screen space pruning, occlusion removal, and feature evaluation are performed to obtain all points that need to be loaded. An irregular triangular mesh is constructed and sent to the GPU for rendering to load the screen. The tree height evaluation criteria are constructed, including: based on the DEM size and the comprehensive index ratio of terrain features, the tree height evaluation criterion Level is constructed. Level = log2W - log2R - h Where W represents the side length of the DEM, R represents the maximum proportion of points selected by the comprehensive index of terrain features when the terrain error is less than the threshold, and h represents the parameter that can be autonomously adjusted in height; a quadtree that meets the tree height evaluation criteria is called a simple quadtree. Constructing land use evaluation criteria, including: Calculate the height of multiple nodes and child nodes to determine the terrain category of the current area, and construct a terrain category quadtree based on a simple quadtree; Take the center point of the node and the center points of its four child nodes after partitioning the node once. If the height of any one of these 5 points is greater than or equal to 200 meters and less than 500 meters, then the area is considered to be hilly terrain and is divided into sections. If the altitude is greater than or equal to 500 meters, the area is considered mountainous and is divided twice. If the distance is less than 200 meters, the area is considered to be a plain and does not need to be divided. In the end, the three types of terrain differ by only 1 in the quadtree, which facilitates subsequent network construction and avoids cracks. Constructing sight distance evaluation criteria, including: First, the maximum effective distance Far is the distance at which the terrain is exactly displayed on the screen: Where W represents the side length of the DEM and Fov represents the camera parameters; Then, the nearest effective distance is defined as the distance at which a certain area of ​​the terrain is exactly displayed on the screen: Near = Far × α Where α represents the scaling factor; Finally, based on the farthest effective distance (Far) and the nearest effective distance (Near), the line-of-sight evaluation criterion Distance is constructed: Where Camera represents the current distance between the camera and the center of observation; Construct vision evaluation criteria, including: The observation area is divided into a central observation area (Inner) and a transition area (Out). Inner = Out × 0.5 Where Out max Out represents the maximum side length of the outer transition region. min Far represents the minimum side length of the outer transition zone, Near represents the farthest effective distance, Distance represents the line-of-sight evaluation criterion. Construct feature evaluation criteria, including: Based on comprehensive terrain feature indicators and sight distance evaluation criteria, a feature evaluation criterion Threshold is constructed: Where F min F represents the minimum comprehensive index of terrain features. max The maximum terrain feature is represented by the comprehensive index, Distance represents the line-of-sight evaluation criterion, Far represents the farthest effective distance, and Near represents the nearest effective distance.

2. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, The topographic factors include slope, surface roughness, surface cutting depth, topographic relief, elevation variation coefficient, profile curvature, and plane curvature.

3. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, The initialization yields a multi-evaluation factor quadtree, including: Based on the digital elevation model (DEM) data, the entire terrain area is used as the root node of the quadtree. It is determined whether the quadtree meets the tree height evaluation criteria. If it does, it is further subdivided; otherwise, it is stopped, resulting in a simple quadtree representing the basic terrain. Then, based on the land use evaluation criteria, the leaf nodes of the simple quadtree were further subdivided to obtain a land use quadtree; Finally, determine whether the leaf nodes of the land type quadtree meet the view evaluation criteria: if they do, it means that the node is in the observation area and needs to be further divided; if they do not meet the criteria, no division is needed, and the viewpoint quadtree is obtained. The construction of the multi-evaluation factor quadtree is completed.

4. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, The process of updating the multi-evaluation factor quadtree includes: after camera parameter changes, first updating the line-of-sight evaluation criterion, then updating the field-of-view evaluation criterion, and determining whether all nodes need to be subdivided or deleted based on the new field-of-sight evaluation criterion; this is divided into the following four cases: 1) Leaf nodes that previously met the field of view evaluation criteria do not need to be updated if they still meet the field of view evaluation criteria after camera parameter changes; 2) If a leaf node that previously met the field of view evaluation criteria no longer meets the criteria after the camera parameters change, it needs to be deleted. At the same time, the parent node of the leaf node is remarked as unsubdivided and becomes a new leaf node. 3) Nodes that previously did not meet the field of view evaluation criteria do not need to be updated if they still do not meet the criteria after camera parameter changes; 4) Nodes that previously did not meet the field of view evaluation criteria, but now meet the criteria after camera parameter changes, need to be subdivided. At the same time, the node is marked as subdivided and four leaf nodes are added.

5. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, Screen space pruning includes: determining whether a leaf node of the multi-evaluation factor quadtree is within the screen; if it is, proceed to the next step; otherwise, skip the node.

6. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, Occlusion removal includes: emitting a ray from the leaf nodes on the screen to the camera position, calculating the position of the ray on the screen, reading the occlusion array at the corresponding position, and determining whether there is an occlusion triangle intersecting the ray. If there is, it means that the occlusion is blocked and no new feature point needs to be added; if not, it means that the occlusion is not blocked. The height of the occlusion triangle is taken as the height of the center point of the terrain tile, and half the size of the terrain tile is taken as the length of the base of the occlusion triangle. The triangle is then added to the occlusion array.

7. The three-dimensional visualization method for multi-resolution digital elevation models according to claim 1, characterized in that, Feature evaluation includes: based on unoccluded leaf nodes, determining whether the points within the range represented by the node meet the feature evaluation criteria. If they do, they are added to the rendering queue; otherwise, they are skipped. Finally, all the required points are obtained, a TIN is constructed, and the TIN is sent to the GPU for rendering.

8. A three-dimensional visualization device for multi-resolution digital elevation models, characterized in that, Including processor and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1 to 7.

Images