CN101694726B - Fusing and drawing method based on multi-source terrain data - Google Patents

Abstract

A fusion and rendering method based on multi-source terrain data, the steps are: (1) Efficient organization of multi-source terrain data (2) Efficient filling of regular "F"-shaped holes at the time of rendering: first obtain high-precision terrain The intersection area of the data and the existing large-scale terrain data, and then according to the clipmap hierarchy, decompose the intersection area into several adjacent rectangular strips of different lengths, and finally correspond to them in the order from fine to rough Insert the corresponding vertices in the strips and ensure the smoothness of the height of the vertices between the strips; (3) draw multi-source terrain data. On the basis of retaining the advantages of the Geometry Clipmap method for drawing large-scale terrain at a high speed, the invention efficiently organizes the imported high-precision terrain data, and especially solves the rules of real-time data fusion caused by clipmap update at the time of drawing. F”-shaped hole problem, while avoiding the jump problem when different resolutions are transitioned during data fusion.

Description

A Fusion and Rendering Method Based on Multi-source Terrain Data

technical field

The invention belongs to the technical fields of computer virtual reality and computer graphics, and in particular relates to a massive terrain rendering method in computer graphics accelerated rendering technology.

Background technique

Large-scale terrain is an important visualization element in the virtual environment. It can provide users with realistic visual experience and is widely used in battlefield simulation, flight simulation, emergency response training and 3D games. In recent years, the rapid development of terrain data acquisition methods has led to a sharp increase in the amount of data, which has brought new challenges to the real-time rendering of large-scale terrain.

There are currently two types of large-scale terrain real-time rendering methods. One is to use out-of-core real-time rendering of large-scale terrain. This type of method effectively organizes large-scale terrain data in external memory, and reduces the amount of loaded data through the scheduling strategy at runtime to complete real-time rendering. It is usually improved on the basis of regular grid-based or irregular grid-based methods. In order to avoid the jump caused by different level of detail (LOD) models in the same area at the time of drawing, a continuous LOD method based on viewpoint is given to solve this problem. In order to solve the problem that frequent I/O operations lead to the reduction of drawing speed, high-efficiency external memory organization and the method of reducing I/O operations through prefetching strategies are used to complete real-time drawing. In order to solve the stability of drawing speed, a smooth scheduling method based on viewpoint is given. Pajarola gave a good summary ("Survey on Semi-Regular Multiresolution Models for Interactive Terrain Rendering", The Visual Computer, 23(8):583-605, 2007). The other is a large-scale real-time rendering method based on memory. This type of method first uses the compression process to reduce the amount of data, so that part or all of the compressed data is loaded into the memory, and secondly synthesizes real data through the decompression operation at runtime and completes real-time rendering. For example, the Geometry Clipmap (GC) method efficiently draws large-scale terrain. This method treats DEM data as a texture composed of elevations, uses a set of nested grids to store and draw terrain; at runtime, the required data is retrieved from the memory. Real-time decompression reduces the time for data to be read from external storage to internal memory; at the same time, the "L"-shaped update strategy is used to avoid extra space usage during the update process and greatly improve the speed of update and drawing. At present, the GC method has become a research hotspot in large-scale terrain rendering, and is used in more and more terrain rendering systems.

With the deepening of the application, people hope that the existing terrain data can provide more detailed changes to achieve specific goals, so as to correctly evaluate and analyze the changed local terrain. The changed details are usually concentrated in some areas of the terrain, resulting in a difference between the sampling accuracy of the local detail data and the sampling accuracy of the surrounding large-scale terrain data, which brings new problems for the organization and update of terrain data. If the detail area is regularly sampled and a hierarchical structure is established according to the regular data organization method based on external memory, when the detail area is large, the depth of the index structure will be increased, thereby increasing the storage space and the traversal overhead at runtime. If the high-precision detailed data is processed according to the GC method, since the terrain is sampled regularly in order to reduce the resolution during preprocessing, once the sampling rate of the local data changes, if the high-precision data is still processed according to the sampling rate of the existing terrain If the detailed data is processed, information loss will occur. In order not to lose details, it is necessary to resample large-scale terrain according to the sampling accuracy of local high-precision detail data, which is expensive and of little significance. Therefore, how to better integrate the high-precision detail data with the existing terrain data and draw them efficiently is the main problem to be solved in this paper.

Contents of the invention

The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a fusion and rendering method based on multi-source topographical data. On the basis of retaining the advantages of the GC method, such as fast rendering speed and small memory occupation, the existing large-scale The terrain data is efficiently organized and drawn, and at the same time, the high-precision terrain data representing the details is well integrated, and the "F"-shaped hole in the data fusion caused by the update of the clipmap structure is effectively solved, and the given transition is used The hole area with better filling structure avoids the problem of smooth transition of different resolution data, forming an effective fusion and rendering method of multi-source terrain data.

The technical solution adopted in the present invention: a method for fusion and rendering based on multi-source terrain data, which is characterized in that the steps are as follows:

(1) Organize multi-source terrain data

First construct the clipmap structure for the existing large-scale terrain data, and then organize the mipmap structure for the high-precision terrain data representing details;

(2) Efficient filling of regular "F"-shaped holes at the drawing moment

Firstly, the intersecting area between the multi-source terrain data and the existing large-scale terrain data is obtained, and then the intersecting area is decomposed into several adjacent rectangular strips (transition strips) of different lengths. Sequentially insert the corresponding vertices in the strip corresponding to the clipmap intersecting layer and use the weighted value method to ensure the smoothness of the height of the vertices between the strips;

(3) Drawing multi-source terrain data

Use the GC method to draw the existing large-scale terrain, use the mipmap structure to draw multi-source data, and use the triangle strip drawing method for the transition zone, and ensure that the addition of the transition zone makes the transition between the areas on both sides of the hole smooth.

The steps of multi-source terrain data organization in described step (1) are as follows:

(1) According to the view point as the center, the clipmap structure of the terrain is constructed with the step size of the formula L=2 ^d as the unit, wherein L represents the step size updated at each moment of the view point, and d is the number of layers of the clipmap;

(2) For the difference between the clipmap data and the real data, use the LBT change to compress it.

The intersecting test steps of the multi-source detail data in the step (2) and the existing large-scale terrain data are as follows:

(1) Update the bounding box of each layer of the clipmap structure;

(2) The bounding box of each layer of the clipmap is intersected with the bounding box of the high-precision terrain;

(3) Record the intersection points in the high-precision terrain, and record the intersection points in each layer of the clipmap at the same time.

The rectangular strip construction step in described step (2) is as follows:

(1) Obtain the intersection point of the largest layer (roughest) in the clipmap intersecting layer and the boundary of high-precision terrain data, and record the upper left and upper right vertices of the intersection boundary;

(2) Obtain the two intersection points on a straight line of the smallest layer (finest) in the clipmap intersecting layer, draw a vertical line to the straight line composed of upper left and upper right vertices, obtain the intersection point, and use this intersection point with high-precision terrain The intersection points on the data boundary together form the transition zone rectangle corresponding to the smallest layer of the clipmap, and the edge formed by the intersection points on the high-precision terrain data boundary in this rectangle is recorded as this transition zone high-resolution edge, and the corresponding other edge is recorded as For this transition band low-resolution edges;

(3) The minimum layer low-resolution edge of the clipmap is used as an input to obtain the transition zone rectangle of the next clipmap layer;

(4) Repeat (3) until the transition zone rectangle of the largest layer of the clipmap is constructed.

The steps of vertex filling in the transition zone of each layer of clipmap in the described step (2) are as follows:

(1) Obtain the smallest layer in the clipmap intersecting layer, copy the vertices on the intersecting edge of this layer to the low-resolution edge of the transition zone of this layer, and copy the vertices on the intersecting edge of the multi-source terrain data corresponding to this layer to The high-resolution edge of the transition zone of this layer;

(2) Obtain the intersecting layer of the next layer of clipmap, copy the vertices on the intersecting edge of this layer of clipmap to the low-resolution edge of the transition zone of this layer, and then compare the adjacent vertices on the low-resolution edge with the corresponding high-precision terrain After linear interpolation on the intersecting edge, press all points into the high-resolution edge of the transition zone of this layer, and then push the vertices in the low-resolution layer in the smallest layer into the high-resolution edge of this layer;

(3) The iterative process (2) until the maximum layer in the clipmap intersecting layer.

The steps of drawing of multi-source terrain data in the described step (3) are as follows:

(1) The GC method draws the existing large-scale terrain, and uses the ring addressing method to reduce redundant copies between data;

(2) Use the mipmap structure based on error control to draw high-precision terrain data representing details;

(3) Draw each transition zone using its own triangulated strip.

Compared with the prior art, the present invention has the beneficial effects that: it provides a real-time method for multi-source terrain data fusion and rendering, which has the characteristics of simple data organization and is conducive to hardware acceleration; and finds that when multi-source data fusion The "F"-shaped hole generated by the clipmap structure update gives a hole repair strategy based on the transition zone; compared with other methods, the transition zone structure has the characteristics of automatic filling according to the size of the hole area during operation, and the filled The vertices ensure the smooth transition of data on both sides of the hole, thus achieving real-time fusion and rendering of multi-source terrain data.

Description of drawings

Fig. 1 is the overall process schematic diagram of the present invention;

Fig. 2 is the schematic diagram of the transition zone structure that the present invention uses;

Fig. 3 is a schematic diagram of vertex filling in the transition zone of the present invention;

Fig. 4 is the schematic diagram of transition zone apex height value of the present invention;

Fig. 5 is the running time LOD selection process of the present invention;

Fig. 6a and Fig. 6b are schematic diagrams showing the effect comparison before and after the use of the present invention.

Detailed ways

Below in conjunction with accompanying drawing and embodiment the present invention is described in further detail:

The implementation process of the invention includes three main steps: multi-source terrain data organization, construction of transition zone and vertex filling, and real-time rendering of multi-source terrain data. As shown in Figure 1.

Step 1 Multi-source terrain data organization:

The present invention uses the clipmap structure to organize the existing large-scale terrain data. In order to better integrate the high-precision detail data and the clipmap structure of the large-scale terrain data at the time of drawing, the mipmap is adopted for the characteristics that the high-precision detail data is a sampling rule. Structures organize high-precision detail data.

The second step is to construct the transition zone and vertex filling, which can be divided into three stages, the formation stage of the transition zone, the determination of the plane coordinates of the filling vertices in the transition zone, and the determination of the height value of the filling vertices in the transition zone:

The first stage, the transition zone formation stage:

In this paper, the "F"-shaped cavity area is divided into several parallel rectangular strips, and each strip is composed of two edges composed of vertices with different sampling rates. Such a band is called the transition zone (Fig. 2). The "F"-shaped cavity is usually composed of multiple transition zones, and the set of transition zones required to fill the "F"-shaped cavity is called a transition zone group. In the transition zone group, several transition zones are connected to each other, and the edge formed by the vertices with high sampling rate in each transition zone is the edge formed by the vertices with low sampling rate in the adjacent zone, so that the data sampling rate is successively reduced or increased in the transition zone group. The high-level process solves the problem of data transition with different sampling precision on both sides of the "F"-shaped hole.

In the second stage, the determination of the plane coordinates of the filled vertices in the transition zone:

Vertices with different sampling rates need to be filled into the two edges of each transition zone. Therefore, the transition zones in the transition zone group should be arranged in order of decreasing or increasing sampling rate. The vertex filling direction used in this paper is from the edge composed of vertices with lower sampling rate on both sides of the "F"-shaped hole to the edge composed of vertices with higher sampling rate. It can be seen from Figure 3 that due to the high sampling accuracy of high-precision detail data, the direction of vertex filling is from the intersection edge of the clipmap structure and the hole area to the intersection edge of the mipmap structure and the hole area;

For the convenience of the following description, record the large-scale terrain data as O, use O _c to represent the c-th layer model in the clipmap structure, and use s(O _c ) to represent the step size of the vertices in this layer model; record the high-precision detail data as D, use D _d represents the dth layer model in the mipmap structure, use s(D _d ) to represent the step size of the vertices in this layer model; record the transition zone data as TS, use TS _i to represent the i-th transition zone, and use TS _i to represent this transition Let s( TS _i ) represent the minimum step size of the vertices in TS _i ; let TS _i represent the edges composed of high-resolution vertices in this transition zone, and use s(TS _i ) to represent TS The minimum step size of vertices in _i ;

Figure 3 contains three transition zones. When filling, first generate TS ₁ , where the vertices of TS ₁ are composed of vertices in the edge where O _m+1 intersects with the hole area in the clipmap structure. The vertex of TS ₁ is composed of the vertex of the vertex in TS ₁ and the vertex of TS ₁ . TS ₂ is generated after the vertex filling of TS ₁ is completed. It can be seen from Figure 3 that the vertex of TS ₂ is composed of the vertex of TS ₁ and the vertex of the edge where O _m intersects the hole area in the clipmap structure. And so on to generate TS ₂ and TS ₃ . During the interpolation process, the sampling rate of the vertices filled by the transition zone i (1≤i≤M) needs to meet the following conditions:

s(TS _i )≥s(D _dmin ) (1)

Considering the extreme case where there are "F" cracks on all four sides of the high-precision detail data and the clipmap structure, the total number of vertices to be filled, totalNumber, satisfies the following relationship:

totalNumber≤(3*2 ⁿ⁺¹ -3n-6)*4*clipsize (2)

The third stage, the determination of the height value of the filling vertex in the transition zone

In order to ensure that the filled vertices and the vertices of the terrain on both sides of the transition zone remain smooth in the height domain, a weighted average method is used to calculate the height value of the newly added vertices (Fig. 4). Let the newly added vertex be V, use h(V) to represent the height value of the vertex, D is the comprehensive factor of the influence of the surrounding vertices on the height of the inserted vertex, wi is the weight of the influence of a certain vertex on the height of the inserted vertex, here use the plane distance instead, then:

$\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; wi</span>}}{\mathrm{<span\; class="notranslate">\; D.</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Vi</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>$ $\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; wi</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

As shown in Figure 4, the newly added vertices on the TS edge of the current strip, v3 and v4 are the vertices on the TS edge in the current transition zone, and v1 and v2 are the vertices on the corresponding high-precision detail area edge. n is the sum of the number of vertices corresponding to the edge intersected by the inserted vertex in the detail area and the border of the clipmap.

Step 3: Multi-source terrain data drawing, which is divided into four stages, the drawing of existing large-scale terrain data, the updating method of high-precision detailed terrain, the LOD selection method of high-precision detailed terrain, and the drawing of transition zone.

The first stage: drawing of existing large-scale terrain data

The method of storing the clipmap data of each layer in the VBO and drawing it is adopted; but at the time of drawing, the area overlapping with the bounding box of the high-precision terrain data in each layer of the clipmap is deleted, and then drawn.

The second stage: rendering of high-precision terrain

Figure 5 shows the update process of the high-precision detail data at runtime. When the center area of the viewpoint is not in the area represented by the high-precision detail data, the mipmap method is used to update and draw the area in real time. However, when the viewpoint is close to the area represented by the high-precision detail data, the amount of updated data of each layer of the mipmap increases with the distance, which affects the rendering speed at runtime.

In order to improve the drawing speed, when the central area of the viewpoint is completely located in the area represented by the high-precision detail data, this paper extracts the clipmap data from the mipmap structure of the high-precision detail data, and uses the "L"-shaped update strategy of the GC method to reduce the amount of data to ensure Real-time rendering of high-precision detail data.

The third stage: LOD selection method for high-precision detailed terrain

In order to reduce the amount of calculation, this paper converts the problem into calculating the projection of the line segment instead of accurately calculating the area projection by calculating the projection of the boundary line segments of each layer of the mipmap structure in the field of view. Let pl _i be the projection of a boundary line segment of a layer of mipmap structure on the screen, nt _i is the number of triangle sides contained in this line segment, and q is the number of boundaries contained in a certain layer of mipmap data in the viewing area. Projection P of mipmap data on the screen in the domain:

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\frac{{\mathrm{<span\; class="notranslate">\; pl</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; nt</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}}{\mathrm{<span\; class="notranslate">\; q</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

The fourth stage: the drawing of the transition zone.

In order to reduce the time overhead brought by the vertex triangulation process in the transition zone, the characteristics of the vertex positions between TS _i and TS _i are used to divide this strip into several connected regions according to the step size of the vertices in TS _i when drawing ( The area formed by the red dotted line segment in Figure 6). And the triangulation of the whole strip is completed by triangulating each region separately.

In order to avoid the triangle loss problem caused by the triangulation of boundary points between regions, the problem is solved by storing the boundary points in the front and back two adjacent regions respectively.

The final rendering result is shown in Figure 6. It can be seen that the introduction of the transition zone structure can effectively solve the fusion and rendering of multi-source terrain data.

Claims (6)

1. A fusion and drawing method based on multi-source terrain data, characterized in that the steps are as follows: (1) Organize multi-source terrain data Multi-source data sources are divided into large-scale terrain data with a single resolution and high-precision terrain data representing details. First, the clipmap structure is constructed for the existing large-scale terrain data in the multi-source terrain data, and then the terrain with high-precision details The data is mipmap structured; (2) Fill the regular "F" shaped hole at the drawing moment Firstly, the intersection area between the high-precision terrain data and the existing large-scale terrain data is obtained, and then the intersection area is decomposed into several adjacent rectangular strips of different lengths. Insert the corresponding vertices into the strips corresponding to the intersecting layer and use the weighted value method to ensure the smoothness of the height of the vertices between the strips; (3) Drawing multi-source terrain data Use the GC method to draw the existing large-scale terrain, use the mipmap structure to draw the high-precision terrain data, and use the triangle strip drawing method for the transition zone, and ensure that the addition of the transition zone makes the transition between the areas on both sides of the hole smooth. 2. the fusion and drawing method based on multi-source terrain data according to claim 1 is characterized in that: in the described step (1), the existing large-scale terrain data in multi-source terrain data is carried out the structure of clipmap structure Proceed as follows: (1.1) According to the viewpoint as the center, the clipmap structure of the terrain is constructed with the step size of the formula L= ^2d as the unit, wherein L represents the step size updated by the viewpoint every moment, and d is the number of layers of the clipmap; (1.2) For the difference between clipmap data and real data, use LBT transformation to compress it. 3. The fusion and drawing method based on multi-source terrain data according to claim 1, characterized in that: the step of obtaining the intersecting area between high-precision terrain data and existing large-scale terrain data in the step (2) is as follows : (2.1) Update the bounding box of each layer of the clipmap structure; (2.2) The bounding box of each layer of the clipmap is intersected with the bounding box of the high-precision terrain; (2.3) Record the intersection points in the high-precision terrain, and record the intersection points in each layer of the clipmap at the same time. 4. the fusion and drawing method based on multi-source terrain data according to claim 1, is characterized in that: the rectangular strip construction step in described step (2) is as follows: (2.1) Obtain the intersection of the roughest layer in the clipmap intersecting layer and the boundary of the high-precision terrain data, and record the upper left and upper right vertices of the intersection boundary; (2.2) Obtain the two intersection points on a straight line of the finest layer in the clipmap intersecting layer, draw a vertical line to the straight line composed of upper left and upper right vertices, obtain the intersection point, and use this intersection point and the boundary of high-precision terrain data The intersecting points above together form the transition zone rectangle corresponding to the finest layer of the clipmap, and the side formed by the intersection points on the high-precision terrain data boundary in this rectangle is recorded as this transition zone high-resolution side, and the corresponding other side is recorded as This transition has low-resolution edges; (2.3) The low-resolution edge of the finest layer of the clipmap is used as an input to obtain the transition zone rectangle of the next clipmap layer; (2.4) Repeat (2.3) until the transition zone rectangle of the roughest layer of the clipmap is constructed. 5. the fusion and rendering method based on multi-source topographical data according to claim 1, is characterized in that: the step of inserting corresponding vertex in described step (2) is as follows: (2.1) Obtain the finest layer in the clipmap intersecting layer, copy the vertices on the intersecting edge of this layer to the low-resolution edge of the transition zone of this layer, and copy the vertices on the intersecting edge of the high-precision terrain data corresponding to this layer in the same way the high-resolution edge of the transition zone to this layer; (2.2) Obtain the intersecting layer of the next clipmap layer, copy the vertices on the intersecting edge of this layer clipmap to the low-resolution edge of the transition zone of this layer, and then compare the adjacent vertices on the low-resolution edge with the corresponding high-precision terrain After the vertices on the intersecting edges are linearly interpolated, all the points are pressed into the high-resolution edge of the transition zone of this layer, and then the vertices in the low-resolution in the finest layer are also pressed into the high-resolution edge of this layer; (2.3) Repeat process (2.2) until the roughest layer in the clipmap intersection layer. 6. the fusion and drawing method based on multi-source terrain data according to claim 1 is characterized in that: the step of drawing of multi-source terrain data in the described step (3) is as follows: (3.1) Use the GC method to draw the existing large-scale terrain, and use the ring addressing method to reduce redundant copies between data; (3.2) Use the mipmap structure based on error control to draw high-precision terrain data.

Images