CN117609524A - Visual analysis method, device and equipment based on three-dimensional R-tree spatial index - Google Patents

Abstract

The present application relates to a visual analysis method, device and equipment based on three-dimensional R-tree spatial index. By establishing a fine-grained 3D-R tree spatial index for each triangular patch in the data set, and then performing a spatial intersection query between the line of sight and the spatial index connecting the viewpoint and the target point, the visibility of the target point relative to the viewpoint is determined. This invention establishes an index database for the triangular patches of the original data, discards the interpolation process of the traditional method, reduces the calculation cost and computational complexity, and can realize rapid visual analysis of massive triangular patch data. In addition, it can also support more The query function can make preliminary preparations for further analysis.

Description

Visual analysis method, device and equipment based on three-dimensional R-tree spatial index

Technical field

The present application relates to the technical field of visibility analysis, and in particular to a visibility analysis method, device and equipment based on a three-dimensional R-tree spatial index.

Background technique

Visibility analysis is an important part of spatial analysis. Two-dimensional plane analysis includes point-to-point visibility analysis and viewpoint-based visual domain analysis. The main research object of visibility is the visibility of the observation point and the observed point, that is, whether one point can be observed from another point, also known as point visibility; the visual field is mainly studied at the given point of observation. The point set of visible points that can be observed within the line of sight is also called surface visibility. The spatial visible point set (isovist) and visual cone model are based on the analysis of three-dimensional spatial models. Especially in urban research, the amount of model data involved in calculations is mostly calculated in the hundreds of millions, which is very huge. Before detailed calculations are carried out, fast and effective two-dimensional spatial analysis has helped the research to do the preliminary screening work, so that in the three-dimensional visual analysis, the viewpoints involved in the calculation are those that perform better in the field of view analysis and have greater potential. of.

Visibility analysis is extended from the basic fact that human eyes observe things. It can directly solve visual problems in real life. It has huge application potential and value: it is currently used in landscape analysis, architectural planning, astronomy and other fields. Very deep application. In landscape analysis or scenic route planning, there are applications such as scenic spots, visiting routes, and infrastructure planning; and the construction and planning of buildings are also inseparable from visibility analysis, which is related to the visual effect of the entire city; in the power industry, The site selection and optimization of communication power towers require visual field analysis to guide the layout; in astronomy, observing planetary objects requires a good ground observation station, and the most important condition is that the visual field effect is the best; in astronomy, observing planetary objects requires a good ground observation station. In many games, it is necessary to take into account the different areas seen by characters in different positions. Therefore, visibility analysis is also an issue that must be considered in the design and development of games.

As the complexity of 3D scenes increases, the use of regular grids to describe ground undulations or simple city models can no longer meet application needs. Instead, oblique photography 3D models are used to describe various complex features including buildings, trees, street lights, billboards, etc. Perform more accurate and realistic spatial expressions. Oblique photography 3D modeling (referred to as oblique 3D) technology is an important restructured modeling technology that has emerged in the field of remote sensing mapping in recent years. Compared with traditional artificial 3D modeling methods, oblique 3D reconstruction has the characteristics of high accuracy and good results. It has the advantages of exquisiteness, low degree of human intervention, short production cycle, and high degree of restoration. Due to the large number of details and high precision restored by tilted 3D models, such 3D models often have a large number of triangle patches and ultra-high texture quality. The amount of data is huge, which increases the difficulty of real-time spatial analysis and visualization. And in the three-dimensional spatial object organization model, these massive triangular patch models are mostly used for GPU visual rendering. In spatial analysis, the corresponding visibility analysis is mostly based on GPU rendering, so the supported spatial analysis capabilities are weak. It is difficult to conduct in-depth mining and analysis of data.

Contents of the invention

Based on this, it is necessary to address the above technical problems and provide a visual analysis method, device and equipment based on three-dimensional R-tree spatial index, so as to achieve rapid visual analysis of massive triangular patch data.

A perspective analysis method based on three-dimensional R-tree spatial index, the method includes:

Obtain the space triangle patch model of the three-dimensional object; the space triangle patch model contains several space triangle patches;

Using each space triangle patch as an index node, construct a three-dimensional R-tree spatial index of the space triangle patch model;

Obtain the line of sight by connecting the current viewpoint and the target point, and perform a spatial intersection query on the line of sight and the three-dimensional R-tree spatial index;

The visibility of the target point is obtained based on the results of the spatial intersection query.

A visual analysis device based on three-dimensional R-tree spatial index, the device includes:

The space triangle patch model acquisition module is used to obtain the space triangle patch model of a three-dimensional object; the space triangle patch model contains several space triangle patches;

The three-dimensional R-tree spatial index building module is used to construct a three-dimensional R-tree spatial index of the spatial triangle patch model using each spatial triangle patch as an index node;

The spatial intersection query module is used to obtain the line of sight from the current viewpoint and the target point, and performs spatial intersection query on the line of sight and the three-dimensional R-tree spatial index;

The target point visibility analysis module is used to obtain the visibility of the target point based on the results of the spatial intersection query.

A computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the following steps:

Obtain the space triangle patch model of the three-dimensional object; the space triangle patch model contains several space triangle patches;

Using each space triangle patch as an index node, construct a three-dimensional R-tree spatial index of the space triangle patch model;

Obtain the line of sight by connecting the current viewpoint and the target point, and perform a spatial intersection query on the line of sight and the three-dimensional R-tree spatial index;

The visibility of the target point is obtained based on the results of the spatial intersection query.

The above-mentioned visual analysis method, device and equipment based on three-dimensional R-tree spatial index establishes a fine-grained 3D-R tree spatial index for each triangular patch in the data set, and then performs the line of sight and spatial index connecting the viewpoint and the target point. Spatial intersection query operation determines the visibility of the target point relative to the viewpoint. Compared with methods such as the view-through algorithm and GPU pipeline rendering in the existing technology, this method establishes an index library for the triangular patches of the original data, discards the interpolation process of the traditional method, reduces the calculation cost and complexity, and can also support More query functions make preliminary preparations for further analysis.

Description of drawings

Figure 1 is a schematic flow chart of a perspective analysis method based on a three-dimensional R-tree spatial index in one embodiment;

Figure 2 is a schematic diagram of the triangular patch of the building model;

Figure 3 is a schematic diagram of fine-grained spatial indexing for spatial triangle patches;

Figure 4 is a schematic diagram of the design process of a fast visibility analysis method based on fine-grained 3D-R tree spatial index;

Figure 5 is a schematic diagram of the parabolic analysis;

Figure 6 is a schematic diagram of the spatial triangle patch index based on 3D-R tree and AABB. Figure 6(a) is the 3D-R tree spatial index structure, and Figure 6(b) is the 3D-R tree;

Figure 7 is a schematic diagram of the 3D-R tree structure that is split after inserting nodes;

Figure 8 is a schematic diagram of the spatial intersection query problem between the line of sight and the triangular patch;

Figure 9 is a schematic diagram of the overall process of visual analysis;

Figure 10 is a schematic diagram of the intersection between the line of sight line segment and the axial bounding box;

Figure 11 is a schematic diagram of the performance of view analysis based on 3D-R tree index;

Figure 12 is a schematic diagram of the time complexity of view analysis based on 3D-R tree index;

Figure 13 is an internal structure diagram of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

In one embodiment, as shown in Figure 1, a perspective analysis method based on a three-dimensional R-tree spatial index is provided, including the following steps:

Step 102: Obtain the spatial triangle patch model of the three-dimensional object.

The space triangle patch model contains several space triangle patches.

Usually, triangular patches directly store coordinate points, and triangular faces are formed from three-dimensional space coordinate points, which are surrounded by triangular patches to form three-dimensional space objects. Triangular patches have the following advantages when expressing objects in the real world:

(1) The data structure is simple, the display speed and lighting calculation are fast, and it is very suitable for parallel processing.

(2) A large number of triangular meshes can be obtained using three-dimensional scanning technology.

(3) When the number of patches is large enough, complex surfaces can be infinitely approximated with arbitrary accuracy.

As shown in Figure 2, a schematic diagram of the triangular patch of the building model is provided.

The structure of triangular patch data is divided into two parts: vertex and face information, which are widely stored in the obj file format, as follows:

(1) Geometric information of vertices: If the triangular patch model contains n vertices, that is, V = {v ₁ , v ₂ ,..., v _n }, v represents the vertex in the obj file, and the lines starting with v are followed by Use three numbers to correspond to the (x, y, z) coordinates of the vertex. The vertex numbers are arranged from small to large.

(2) Represent the topological connection relationship between triangular meshes through triangle sides and triangular faces. Every two vertices represent the sides of a triangle, such as (v ₁ , v ₂ ). Every three vertices in the counterclockwise direction represent the corresponding surface, such as (v ₁ , v ₂ , v ₃ ). Since the sides of a triangle can be found from the faces of the triangle, the triangle patch model only needs to contain information about the triangle patches. If the triangular patch model contains m faces, then F={f ₁ , f ₂ ,..., f _m }. In the triangle patch model file, f represents the triangle patch. Each line uses f as the starting mark. After f are three integers, corresponding to the three vertex numbers of the triangle patch, arranged in counterclockwise order.

Step 104: Use each space triangle patch as an index node to construct a three-dimensional R-tree spatial index of the space triangle patch model.

For data division and spatial index construction of large-scale geographical vector elements, commonly used methods include object segmentation and rule segmentation. Among them, the object segmentation method is implemented by using hierarchical bounding bodies. Commonly used bounding bodies include: bounding spheres (Spheres). ), Axis-aligned Bounding Box (ABB), Oriented Bounding Box (OBB), Discrete Orientation Polytopes (DOP) and Convex Hull. The object segmentation algorithm requires calculation, segmentation, intersection and other judgments of the bounding body. As the compactness of the bounding body increases, the corresponding computational complexity becomes higher and higher. The rule segmentation algorithms include regular grids, quadtrees/octrees, KD trees, BSP trees and R-trees, etc. Traditional methods directly segment the entire geographical vector elements, resulting in the loss of local detailed information, and are not suitable for accurate visibility analysis.

The triangular patch model has a huge number of patches. When performing visual calculations on hundreds of millions of triangular patches, the intersection operation takes a lot of time and is inefficient. Therefore, this method proposes to extend the R-tree that is commonly used in two-dimensional space to three-dimensional space, establish a spatial data structure of 3D-R tree, and perform perspective calculation of triangular patches based on this data structure. This algorithm inherits the good spatial search ability of R-tree and adopts a bottom-up construction method to reduce the overlapping range of the minimum boundary rectangle of each node while ensuring the balance of the tree. Extend the R tree to three-dimensional space so that it can well establish spatial indexes for triangular patches. As shown in Figure 3, a schematic diagram of fine-grained spatial indexing for spatial triangular patches is provided. Each triangular patch is used as an index node to construct a fine-grained 3D-R tree spatial index for three-dimensional objects. Among them, box represents the bounding box, Point(X _Amin , Y _Amin , Z _Amin ) and Point(X _Amax , Y _Amax , Z _Amax ) respectively represent the minimum coordinate point and the maximum coordinate point of the bounding box corresponding to the triangle patch A.

The leaf nodes of the R tree contain specific data objects, and when performing spatial indexing, the root node of the tree is gradually searched according to the numerical range until the leaf node to find the target data object (in this method, the target data object is Refers to triangular patches), this method significantly improves the retrieval efficiency.

Step 106: Obtain the line of sight by connecting the current viewpoint and the target point, and perform a spatial intersection query on the line of sight and the three-dimensional R-tree spatial index.

After the 3D-R tree index is successfully constructed, the visual analysis problem can be converted into the spatial intersection query problem of the visual line and the triangular patch connecting the viewpoint and the target observation point.

Step 108: Obtain the visibility of the target point according to the result of the spatial intersection query.

As shown in Figure 4, a schematic diagram of the design process of a fast visibility analysis method based on fine-grained 3D-R tree spatial index is provided.

It is worth noting that, in addition to being applied to the visibility analysis of straight line segments between two points, this method can also be applied to the visibility analysis between curves. As shown in Figure 5, a schematic diagram of a parabola-oriented comprehensive analysis is provided. First, the parabola is discretely disassembled into several straight line segments. These straight line segments can be approximately fitted into a parabola curve, and then each discrete straight line segment is disassembled in turn. This method is used to perform visibility analysis and query to obtain the visibility of each straight line segment. If an intersection occurs during the query process, the entire parabola is judged to be invisible.

In the above-mentioned view analysis method based on the three-dimensional R-tree spatial index, a fine-grained 3D-R tree spatial index is established for each triangular patch in the data set, and then the line of sight and the spatial index connecting the viewpoint and the target point are subjected to a spatial intersection query operation. , determine the visibility of the target point relative to the viewpoint. This method can also use multi-threaded parallel computing in the future to further speed up the visual analysis of massive triangular patch data. Compared with methods such as the view-through algorithm and GPU pipeline rendering in the existing technology, this method establishes an index library for the triangular patches of the original data, discards the interpolation process of the traditional method, reduces the calculation cost and complexity, and can also support More query functions make preliminary preparations for further analysis.

It should be understood that although various steps in the flowchart of FIG. 1 are shown in sequence as indicated by arrows, these steps are not necessarily executed in the order indicated by arrows. Unless explicitly stated in this article, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in Figure 1 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution of these sub-steps or stages The sequence is not necessarily sequential, but may be performed in turn or alternately with other steps or sub-steps of other steps or at least part of the stages.

R-trees often use the concept of Minimal Bounding Rectangle (MBR) to surround data items. In two-dimensional space, MBR appears as a rectangle, while in three-dimensional space, it evolves into a cuboid. Starting from the leaf nodes, MBR is used to surround the data items, and the parent node also uses MBR to frame the MBR of its child nodes. As the level increases, the space covered by MBR gradually expands. Finally, the MBR of the root node encloses the entire data. set included. The node structure of an n-order R-tree can be defined as follows:

Leaf node: (Count,Level,<OI ₁ ,MBR ₁ >,<OI ₂ ,MBR ₂ >…<OI _n ,MBR _n >);

Intermediate nodes: (Count,Level,<CP ₁ ,MBR ₁ >,<CP ₂ ,MBR ₂ >…<CP _n ,MBR _n >);

Among them, <OI _n ,MBR _n > represents the target data object, OI _n is the identification of the space target, and MBR _n is the minimum bounding rectangle of the target; <CP _n ,MBR _n > represents the data index, and CP _n points to the subtree root. Pointer to the node. Count≤m indicates the number of data items or index items used in the node, and Level indicates the number of levels of the node in the tree.

The invention mainly includes four data structures: minimum axial bounding box, line of sight structure, 3D-R tree node, and 3D-R tree. And introduce the main member variables:

(1) Minimum axial bounding box: In the case of R-tree in three dimensions, the minimum bounding rectangle of each data item evolves into the form of an axial bounding box, using the Max-Min representation, that is, the axial bounding box is in three directions the maximum and minimum values on.

struct Rect3d{

double m_Min[3];

double m_Max[3];

};

(2) Line of sight structure: The spatial positions of the viewpoint and target point constitute the line of sight structure.

struct Seg3d{

s_start[3];

s_end[3];

};

(3) 3D-R tree node: mainly includes child node pointer, pointer to the next node, as well as data ID and level index.

struct Node{

m_child*;

m_next*;

m_data;

m_level;

};

(4) 3D-R tree: Establish an index structure for triangular patches, which mainly includes node information, axial bounding box information and line of sight structure information.

class 3d_Rtree{

struct Node;

struct Rect3d;

struct Seg3d;

};.

In one embodiment, each space triangle patch is used as an index node to construct a three-dimensional R-tree spatial index of the space triangle patch model, including:

Get the root node of the corresponding layer based on the current triangle patch;

Traverse the child nodes of the root node and find the child node that causes the smallest expansion of the minimum boundary cuboid due to the insertion of the current triangle patch as the target child node;

After inserting the current triangular patch into the corresponding target sub-node, if the number of triangular patches in the target sub-node does not exceed the preset threshold of the corresponding target sub-node, then there is no need to split the target sub-node and a new The target child node;

If the new target sub-node is the root node of its layer, the operation stops and the insertion of the current triangular patch is completed.

In one embodiment, the method further includes:

When there are multiple child nodes with the smallest expansion degree of the minimum boundary cuboid due to the insertion of triangular patches, select the child node with the smallest volume of the minimum boundary cuboid as the target child node.

In one embodiment, the method further includes:

After inserting the current triangular patch into the corresponding target sub-node, if the number of triangular patches in the target sub-node exceeds the preset threshold of the corresponding target sub-node, the target sub-node will be split into the first sub-node and the third sub-node. Two child nodes;

If the first child node is not the root node of its layer, find its parent node according to the level index, and adjust the minimum boundary cuboid of the parent node so that the adjusted minimum boundary cuboid can surround the first child node and the second child node;

If the adjusted minimum boundary cuboid exceeds the storage constraints of the parent node, the parent node is split to obtain the first parent node and the second parent node;

If the first parent node is the root node of the layer where it is located, return to the first parent node and stop the operation; if the first parent node is not the root node of the layer where it is located, find its parent node according to the level index and adjust the parent node's The minimum boundary cuboid enables the adjusted minimum boundary cuboid to surround the first parent node and the second parent node.

In one embodiment, after the target sub-node is split into a first sub-node and a second sub-node, the triangular patch in the original target sub-node and the current triangular patch are inserted into the first sub-node respectively according to the nearest neighbor strategy. node and the second child node.

In order to construct the 3D-R tree spatial index structure, the triangular patch data is inserted in sequence. The nodes of the R tree are dynamically split according to the set threshold, and the height of the R tree is adjusted accordingly. The construction process involves data insertion, minimum expansion node query, and uploading of splitting information, and finally generates a complete three-dimensional surface tree. As shown in Table 1, the construction process of 3D-R tree is provided.

Table 1 3D-R tree construction process

In order to reflect the above operation more intuitively, the process of inserting a triangular patch into the 3D-R tree will be explained through diagrams. The core idea of the algorithm of the present invention is to construct an axial bounding box for each spatial triangle patch and build a 3D-R tree from bottom to top. As shown in Figure 6, a schematic diagram of the spatial triangle patch index based on 3D-R trees and AABB is provided, in which Figure 6(a) is the 3D-R tree spatial index structure, and Figure 6(b) is the 3D-R tree. p ₁ , p ₂ ,..., p ₅ represent the spatial triangle patch of the original data, and the solid line cube represents the axial bounding box of the spatial triangle patch, using the minimum (X _min , Y _min , Z _min ) and the maximum ( X _max , Y _max , Z _max ) coordinate points are represented. According to the principle of R tree, continue to integrate upward to obtain the index space R ₁ and R ₂ corresponding to the root node. According to the constructed spatial index library, the axial bounding box of each triangular patch can be quickly queried, thereby quickly obtaining the spatial index data that intersects the line of sight. When inserting a new triangular patch P ₆ , first query the minimum expansion node. There are two AABBs in the root node, namely R ₁ and R _2. Comparing the two, it can be seen that R ₁ has been partially surrounded P ₆ is removed, and if inserted into R ₂ , the Range of R ₂ will increase a lot. Therefore R ₁ is selected as the insertion node. However, since the threshold set in the tree is 3, that is, the maximum object that a node can store is 3, the inserted node needs to be split.

Split the root node R ₁ into R ₁ and R ₁ ', and at the same time choose to put P ₂ and P ₆ into one node, P ₁ and P ₃ into another node according to the nearest neighbor strategy, and propagate upward. Since there are only It contains R ₁ and R ₂ and does not exceed the threshold, so there is no need to split. You only need to adjust the Range to ensure that the leaf nodes are surrounded. The split 3D-R tree structure after inserting the node is shown in Figure 7.

In one embodiment, performing a spatial intersection query on the line of sight and the three-dimensional R-tree spatial index includes:

Get the starting point coordinates and end point coordinates of the line of sight;

Find the minimum boundary cuboid of the corresponding node from the three-dimensional R-tree spatial index according to the starting point coordinates and the end point coordinates;

Determine whether the line of sight intersects the minimum boundary cuboid of the corresponding node based on the starting point coordinates and end point coordinates;

If they intersect, further obtain the triangular patch information in the corresponding node;

Determine whether the line of sight intersects the triangle patch in the corresponding node based on the triangle patch information.

In one embodiment, obtaining the visibility of the target point based on the results of the spatial intersection query includes:

If the line of sight does not intersect with the minimum boundary cuboid of the corresponding node, the target point is visible;

If the line of sight intersects the minimum boundary cuboid of the corresponding node but does not intersect with the triangular patch in the corresponding node, the target point is visible;

If the line of sight intersects the minimum boundary cuboid of the corresponding node and the triangle patch in the corresponding node, the target point is invisible.

As shown in Figure 8, a schematic diagram of the spatial intersection query problem between the line of sight and the triangular patch is provided. First, the intersection test of the line of sight and the axial bounding box is performed, and then the intersection test of the line segment and the triangular patch is performed to obtain the visibility of the target point. .

As shown in Figure 9, a schematic diagram of the overall process of visual analysis is provided. After the construction of the 3D-R tree spatial index is completed, the spatial intersection query stage is entered. When performing a line of sight analysis between two points, it is necessary to perform a spatial intersection query on the line connecting the viewpoint and the target point, that is, the line of sight and the 3D-R tree. If they intersect, the two points do not see each other. If they do not intersect, it means The target point is visible. The above problem can be converted into a mathematical geometry problem, that is, the spatial intersection problem of a straight line segment and an axial bounding box, and the further analysis of the intersection problem of a straight line segment and a triangle.

As shown in Figure 10, a schematic diagram of the intersection of the line of sight line segment and the axial bounding box is provided.

Let the starting point and end point of the line segment be respectively The maximum and minimum points of the axial bounding box AABB are A (X _max , Y _max , Z _max ) and B (X _min , Y _min , Z _min ) respectively. Any point on the line segment can be expressed as:

in represents the direction of the line segment, and t represents the diagonal distance from the starting point to any point. To determine whether a line segment intersects a cube, the core principle is that the part of the line segment that passes through the cube is the intersection of the parts of the six faces that intersect with two planes. The problem can be converted into the maximum oblique distance of the line segment into the plane max(t _{x_near} ,t The relationship between _{y_near} ,t _{z_near} ) and the minimum oblique distance min(t _{x_far} ,t _{y_far} ,t _{z_far} ) from the plane. If the maximum oblique distance entering the plane is less than the minimum oblique distance leaving the plane, it means that the line segment is Intersection of cubes:

max(t _{x_near} ,t _{y_near} ,t _{z_near} ,0)≤min(t _{x_far} ,t _{y_far} ,t _{z_far} ,1) (2)

Taking X-slab as an example, the intersection point of the line segment and the X plane is:

x _start +t*d _x =X _min (3)

Just solve the coefficient t. In the same way, you can find the coefficient t of Y-slab and Z-slab, and then use the judgment conditions of Formula 2 to get the Boolean operation result. As shown in Table 2, the pseudo code of the spatial intersection query algorithm process of the straight line segment and the axial bounding box AABB is provided.

Table 2 Spatial intersection query algorithm flow of straight line segment and axial bounding box AABB

In order to improve the accuracy of the judgment results, this method needs to further determine the spatial relationship between the current line of sight and the triangular patch inside the current axial bounding box after completing the spatial intersection query between the line of sight and the 3D-R tree. A basic detection algorithm It calculates the intersection point between the target object and the plane where the triangle is located, and then determines whether the intersection point is within the triangle. If the intersection point is within the triangle, the object intersects the triangle; otherwise, it does not intersect. Since floating point numbers in computers occupy limited bytes, the accuracy of representation is limited. When calculating the intersection of a straight line and a plane, multiple multiplication or division operations on the same variable will lead to error accumulation, and the final calculation result may be larger due to accuracy issues. Large errors cause the above algorithm to have low efficiency and prone to floating point errors.

This method can quickly obtain the coordinates of the intersection point and center of gravity through vector and matrix calculations without precomputing the plane equations containing triangles.

Assume that the three vertices of the triangle are P _o , P ₁ , and P ₂ respectively. When the line segment intersects the triangle, we can get:

Among them, (u, v)∈Ω = {(u, v): u ≥ 0, v ≥ 0, u + v ≤ 1} is called the center of gravity coordinate, which is determined by solving the range limits of the three variables t, u, v Whether they intersect or not, the above formula can be solved using Cram's rule:

Each vector parameter is defined as:

If the obtained solution (u, v) ∈ Ω and t ∈ [0, 1], then the intersection point of the line and the plane is located inside the triangle; otherwise, it is located outside the triangle. If u=0 or v=0 or u+v=1, the intersection point is on the side of the triangle.

As shown in Table 3, the pseudo code of the intersection algorithm of straight lines and triangle patches is provided.

Table 3 Pseudo code of intersection algorithm of straight line and triangle patch

In order to analyze the visibility performance of the algorithm of the present invention on massive data sets, triangular patch data sets of different sizes in area A were selected for test operations, and the visibility performance and time complexity of the algorithm under different data set sizes were analyzed. The specific configuration of the algorithm testing platform of the present invention is shown in Table 4.

Table 4 Visual analysis performance experimental environment configuration

Test and analysis were carried out on data scales of 100,000, 1 million, 10 million and 100 million units of triangular patch data respectively, a rectangular bounding box was constructed for each triangular patch, and a 3D-R tree fine-grained spatial index was generated. Select an appropriate viewpoint position within each data set, and perform visual analysis on each pixel within the data range. The visual performance of different data scales is shown in Table 5. As the amount of data increases exponentially, visual analysis The performance does not decrease linearly, and its floating control is within 100,000 times per unit time.

Table 5 Visual analysis experimental results

As shown in Figure 11, a schematic diagram of the through-view analysis performance based on 3D-R tree index is provided. When the number of triangular patches is less than 100 million, the intersection operation between the line of sight and the fine-grained spatial index of the 3D-R tree can reach up to 380,000 times per unit time, and slowly decreases as the size of the data set increases. When the number of triangular patches reaches more than 100 million, its visual analysis performance per unit time gradually converges and becomes stable. This shows that the visual analysis performance of the algorithm of the present invention is less affected by the size of the data set. In terms of time complexity, as shown in Figure 12, a schematic diagram of the time complexity of view analysis based on 3D-R tree index is provided. When the amount of triangular patch data is less than 400 million, the time complexity increases logarithmically, that is, O(log _k n), where k is the maximum number of node splits. However, after the magnitude of triangular patches reaches more than 400 million, the calculation time gradually increases. This is because as the size of the data increases, the spatial intersection query operation becomes more complex. When the visibility performance gradually converges, the query time is bound to increase. There will be a growth trend. The above test operations show the performance analysis results under single-threaded operation. Multi-threaded operation will be enabled in the future, which can greatly improve computing performance.

Compared with the traditional sight-through analysis method, the algorithm of the present invention does not need to interpolate the elevation of terrain points. Instead, it establishes a spatial index on the triangular patch, and directly calculates whether the sight line intersects with the terrain in the already constructed index. Searching in the library greatly reduces the complexity of the algorithm, and when tested and analyzed on massive data sets, its visual analysis performance gradually converges.

In summary, as the complexity of three-dimensional scenes continues to increase and GIS data becomes increasingly enriched, we are faced with long-standing problems when conducting visual analysis for a wide range of three-dimensional scenes, including the usability, operability, and security of visual analysis. These problems seriously restrict the practical application of visual analysis. This method conducts in-depth research on the problem of fast view analysis in large-scale scenes, and proposes a fast view analysis method based on 3D-R tree index. It establishes a 3D-R tree fine-grained spatial index for the triangular patch model, and sequentially The connection between the viewpoint and the target point, that is, the line of sight and the spatial index, performs spatial intersection query operations to avoid the height interpolation process of the traditional algorithm, and can quickly and efficiently complete visual analysis on data sets of different sizes. The single-threaded computing performance can reach up to 1 second. 380,000 times. Compared with traditional algorithms, the time complexity of R ² , R ³ and Xdraw algorithms increases with the quadratic or even cubic power of the field of view radius, which is not suitable for large-scale view field analysis problems under massive data sets. , and the time complexity of this method is logarithmic time, which can support long-distance visibility analysis and provide new ideas for subsequent visibility application problems based on large-scale data sets.

In one embodiment, a visual analysis device based on a three-dimensional R-tree spatial index is provided, including: a spatial triangle patch model acquisition module, a three-dimensional R-tree spatial index construction module, a spatial intersection query module and target point visibility Analysis module, which:

The space triangle patch model acquisition module is used to obtain the space triangle patch model of a three-dimensional object; the space triangle patch model contains several space triangle patches;

The three-dimensional R-tree spatial index building module is used to construct a three-dimensional R-tree spatial index of the spatial triangle patch model using each spatial triangle patch as an index node;

The spatial intersection query module is used to obtain the line of sight from the current viewpoint and the target point, and performs spatial intersection query on the line of sight and the three-dimensional R-tree spatial index;

The target point visibility analysis module is used to obtain the visibility of the target point based on the results of the spatial intersection query.

Regarding the specific limitations of the visual analysis device based on the three-dimensional R-tree spatial index, please refer to the above-mentioned limitations on the visual analysis method based on the three-dimensional R-tree spatial index, which will not be described again here. Each module in the above-mentioned three-dimensional R-tree spatial index-based perspective analysis device can be implemented in whole or in part by software, hardware, and combinations thereof. Each of the above modules may be embedded in or independent of the processor of the computer device in the form of hardware, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in Figure 13. The computer device includes a processor, memory, network interface, and database connected through a system bus. Wherein, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage medium stores operating systems, computer programs and databases. This internal memory provides an environment for the execution of operating systems and computer programs in non-volatile storage media. The database of the computer device is used to store data such as triangle patches. The network interface of the computer device is used to communicate with external terminals through a network connection. The computer program, when executed by a processor, implements a perspective analysis method based on a three-dimensional R-tree spatial index.

Those skilled in the art can understand that the structure shown in Figure 13 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Specific computer equipment can May include more or fewer parts than shown, or combine certain parts, or have a different arrangement of parts.

In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps of the method in the above embodiment are implemented.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage. In the media, when executed, the computer program may include the processes of the above method embodiments. Any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The technical features of the above embodiments can be combined in any way. To simplify the description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, all possible combinations should be used. It is considered to be within the scope of this manual.

The above-described embodiments only express several implementation modes of the present application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all fall within the protection scope of the present application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims (10)

1. A method for visual analysis based on three-dimensional R-tree spatial index, the method comprising:

acquiring a space triangular patch model of a three-dimensional object; the space triangular surface patch model comprises a plurality of space triangular surface patches;

constructing a three-dimensional R tree space index of the space triangular patch model by taking each space triangular patch as an index node;

obtaining a sight line by connecting a current viewpoint with a target point, and carrying out space intersection query on the sight line and the three-dimensional R tree space index;

and obtaining the visibility of the target point according to the result of the space intersection query.

2. The method of claim 1, wherein constructing a three-dimensional R-tree spatial index for the spatial triangular patch model with each spatial triangular patch as an inode comprises:

acquiring a root node of a corresponding layer according to the current triangular patch;

traversing the child nodes of the root node, and finding out the child node with the minimum boundary cuboid expansion degree caused by the insertion of the current triangular patch as a target child node;

after inserting the current triangular patches into the corresponding target sub-nodes, if the number of the triangular patches in the target sub-nodes does not exceed the preset threshold value of the corresponding target sub-nodes at the moment, splitting the target sub-nodes is not needed, and a new target sub-node can be obtained;

and if the new target child node is the root node of the layer where the new target child node is located, stopping the operation and completing the insertion of the current triangular patch.

3. The method according to claim 2, wherein the method further comprises: when there are a plurality of child nodes having the smallest expansion degree of the smallest bounding box due to the insertion of the triangular patches, a child node having the smallest volume of the smallest bounding box is selected as the target child node.

4. The method according to claim 2, wherein the method further comprises:

after inserting the current triangular patches into the corresponding target sub-nodes, if the number of the triangular patches in the target sub-nodes exceeds a preset threshold value of the corresponding target sub-nodes at the moment, splitting the target sub-nodes into a first sub-node and a second sub-node;

if the first child node is not the root node of the layer where the first child node is located, finding a father node according to the hierarchy index, and adjusting the minimum boundary cuboid of the father node so that the adjusted minimum boundary cuboid can surround the first child node and the second child node;

if the adjusted minimum boundary cuboid exceeds the storage constraint of the father node, splitting the father node to obtain a first father node and a second father node;

if the first father node is the root node of the layer where the first father node is located, returning to the first father node, and stopping operation; if the first father node is not the root node of the layer, finding the father node according to the hierarchical index, and adjusting the minimum boundary cuboid of the father node, so that the adjusted minimum boundary cuboid can surround the first father node and the second father node.

5. The method of claim 4, wherein after splitting the target child node into a first child node and a second child node, the triangular patches and the current triangular patches in the original target child node are inserted into the first child node and the second child node according to a nearest neighbor policy, respectively.

6. The method of claim 1, wherein spatially intersecting the line of sight and the three-dimensional R-tree spatial index comprises:

acquiring a starting point coordinate and an end point coordinate of the sight;

finding the minimum boundary cuboid of the corresponding node from the three-dimensional R tree space index according to the starting point coordinate and the ending point coordinate;

judging whether the sight line intersects with a minimum boundary cuboid of a corresponding node according to the starting point coordinate and the ending point coordinate;

if the triangular patches are intersected, triangular patch information in the corresponding nodes is further acquired;

and judging whether the sight line intersects with the triangular patch in the corresponding node according to the triangular patch information.

7. The method of claim 1, wherein deriving the visibility of the target point from the results of the spatial cross-query comprises:

if the sight line is not intersected with the minimum boundary cuboid of the corresponding node, the target point is visible;

if the line of sight intersects a minimum bounding cuboid of the corresponding node, but does not intersect a triangular patch in the corresponding node, the target point is visible;

and if the sight line intersects with the minimum boundary cuboid of the corresponding node and intersects with the triangular patch in the corresponding node, the target point is invisible.

8. A three-dimensional R-tree spatial index-based visualization analysis apparatus, the apparatus comprising:

the space triangular patch model acquisition module is used for acquiring a space triangular patch model of the three-dimensional object; the space triangular surface patch model comprises a plurality of space triangular surface patches;

the three-dimensional R tree space index construction module is used for constructing a three-dimensional R tree space index of the space triangular patch model by taking each space triangular patch as an index node;

the space intersection query module is used for obtaining a sight line from a current viewpoint and a target point, and performing space intersection query on the sight line and the three-dimensional R tree space index;

and the target point visibility analysis module is used for obtaining the visibility of the target point according to the result of the space intersection query.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 7 when the computer program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.