KR100724609B1 - A method for generating a cube mesh structure for a terrain rendering - Google Patents

Abstract

A method for generating a cube mesh structure for terrain rendering is provided to solve problems of conventional methods for accelerating voxel-based ray casting and enable effective terrain rendering. The standard deviation for all of height field data which represents a height value of terrain is calculated and the standard deviation is multiplied by a constant to obtain a division reference value(S01). All the height field data is grouped into a block and the block is divided(S02). A difference between maximum and minimum values of the height field data is obtained for the block(S03). The difference between the maximum and minimum values is compared with the division reference value(S04). When the difference is greater than the division reference value, the block is divided into four(S05). When the difference is smaller than the division reference value, the division is finished and the maximum and minimum values and position coordinates are stored(S06).

Description

A METHOD FOR GENERATING A CUBE MESH STRUCTURE FOR A TERRAIN RENDERING}

1 illustrates a method for selecting meaningful voxels (contributing voxels) in polygonal assisted ray projection (PARC).

FIG. 2 is a diagram illustrating an envelope method, and FIGS. 2A and 2B are diagrams illustrating a method of generating an envelope mesh in an envelope method, and a convex problem that may occur in the envelope method. Drawing through the surrounding mesh).

Figure 3 is a block diagram showing the steps of the cube mesh generating method according to an embodiment of the present invention.

4 illustrates a method of dividing an entire block of altitude data using altitude data in accordance with an embodiment of the present invention.

FIG. 5 is a diagram illustrating a process of generating a cube mesh according to an embodiment of the present invention. FIG. 5A is a diagram illustrating maximum and minimum value information for each block obtained when partitioning of blocks is completed. 5B is a diagram illustrating the generation of a cube using the maximum value information of FIG. 5A, and FIG. 5C illustrates a final cube mesh by resizing and reflecting the minimum value information in the cube generated in FIG. 5B and combining them. Drawing showing that.

6 illustrates an example of a cube mesh finally created according to the present invention.

7 is a view comparing the rendering time (time of the rendering step except the preprocessing step) of the rendering method and the envelope method using the cube mesh structure generated according to the present invention.

8 is a view comparing the resultant image of the method according to the present invention and the envelope method, Figures 8a and 8b is a view showing the resultant image generated by the envelope method, Figure 8c is a cube mesh according to the present invention A diagram showing a resultant image generated using.

<Explanation of symbols for main parts of the drawings>

310: altitude data

320: processing information required for quad tree generation

330: step of creating a quad tree

340: step of creating a cube mesh

S01: step of calculating a split reference value

S02: step of starting dividing the block using all the altitude data as one block

S03: calculating a difference between the maximum value and the minimum value of the altitude data for the block

S04: comparing the difference between the maximum value and the minimum value with the divided reference value

S05: divide the block into quarters

S06: ending the split for the block

S06_2: storing maximum, minimum and position coordinates of the altitude data for the block

S07: checking whether a block capable of splitting remains

S08: repeating steps S03 to S07 for the block

S09: end of division for all blocks

S10: step of creating a real cube in a three-dimensional space

The present invention relates to a method for generating a cube mesh structure for terrain rendering. More specifically, the difference between the maximum value and the minimum value of the elevation data of a specific block is divided based on the standard deviation of the elevation data. The voxel-based ray projection method determines whether a specific block is divided by a method of comparing with a reference value, and generates a cube mesh structure that surrounds the terrain in three-dimensional space by using the information of the blocks in which the split is finally completed. The present invention relates to a method of generating a cube mesh structure that solves the problems of existing methods for speeding up a graph while simultaneously enabling effective terrain rendering.

Terrain rendering is used in various fields such as Geographic Information System (GIS), flight simulation, interactive computer games and the like. Triangular meshes and other polygonal models are commonly used for terrain rendering, but height field data is also commonly used to represent terrain. Ray casting is a method of rendering terrain using elevation data, and is characterized by directly rendering a scene without geometric modeling such as a triangle irregular network (TIN). Ray projection has the advantage of producing realistic and good images, but also has the disadvantage that it takes much time to execute due to the inherent complexity of the algorithm. For example, previously studied voxel-based ray projection methods require a large amount of CPU calculation to perform visual frustum screening or step selection.

Various studies have been conducted to reduce CPU computation in voxel-based ray projection. Polygon Assist Ray Casting (PARC) is one such study. FIG. 1 is a diagram illustrating a method of selecting meaningful voxels (contributing voxels) in polygonal assisted ray projection (PARC). As shown in FIG. 1, PARC determines a pair of polygons surrounded by the distance between the closest and furthest voxels that affect (contribute) the last image along each ray and use it for rendering. to be. Typically, these distance values are obtained by projection of the surrounding polygonal model. The original PARC algorithm was a method in which progressive image sharpening was performed using time coherence. In the first frame, the PARC algorithm has to perform calculations on the surrounding polygons for the entire voxel data, so rendering speed is relatively slow. However, as time goes by, the surrounding polygon becomes denser than the previous frame, and the rendering speed is gradually increased because only the voxels affecting the final image are considered.

The first PARC algorithm was designed to accelerate volume ray projection. Similarly, recently, the envelope method using ray projection, which can be applied to elevation data, has been announced. FIG. 2 is a diagram illustrating an envelope method, and FIGS. 2A and 2B are diagrams illustrating a method of generating an envelope mesh in an envelope method, and a convex problem that may occur in the envelope method. (The problem of passing through the surrounding mesh). The envelope method uses graphics hardware and extends the voxel-based terrain rendering technique, which creates an enclosed polygon mesh (envelope mesh) that surrounds the surface of the terrain during the rendering phase. In the envelope method of the example shown in FIG. 2A, after dividing the total altitude data into blocks of size n × n, selecting points having the maximum value of each block, triangular meshes are generated by connecting these points. However, the envelope method, as shown in Fig. 2b, generates a surrounding polygon by considering only the maximum value for each block, which can cause a very fatal problem (convex problem) in which the actual terrain penetrates the surrounding polygon. This will cause distortion in the final image. One way to solve the convex problem in the envelope method is to reduce the size (n) of the dividing block. However, according to the applicant's actual experiment, the convex problem was not easily solved even though the size of the block was sufficiently reduced. In summary, the envelope method can solve some of the convex problem by making the block size very small (n), but in return, in the preprocessing and rendering steps, It takes a lot of time to triangulate a large number of vertices. Therefore, while maintaining the advantages of the envelope method of accelerating voxel-based ray projection, a new method is needed to solve the convex problem.

The present invention originates from the above recognition of necessity, and determines whether to partition the specific block by a method of comparing the difference between the maximum value and the minimum value of the elevation data of the specific block with the division reference value based on the standard deviation of the entire elevation data. By creating a cube mesh structure that surrounds the terrain in three-dimensional space using the information of the blocks whose segmentation is finally finished, while solving the problems of existing methods for accelerating the voxel-based ray projection method, Its purpose is to provide a method for generating a cube mesh structure that enables effective terrain rendering.

Cube mesh structure generation method that can be used for terrain rendering according to the characteristics of the present invention for achieving the above object,

(1) obtaining a standard deviation of the whole elevation data using height field data indicating elevation values of the terrain, and then multiplying the obtained standard deviation by a constant value to obtain a split reference value;

(2) starting dividing the block using the entire altitude data as one block;

(3) calculating a difference between the maximum value and the minimum value of the altitude data for the block;

(4) comparing the difference between the obtained maximum and minimum values of the altitude data for the corresponding block with the split reference value;

(5) dividing the corresponding block into quarters if the difference between the maximum value and the minimum value of the altitude data for the corresponding block is greater than the division reference value as a result of the comparison in the step (4);

(6) If the difference between the maximum value and the minimum value of the altitude data for the corresponding block is less than or equal to the division reference value, as a result of the comparison in the step (4), the division of the corresponding block is terminated, and Storing a maximum value, a minimum value, and a position coordinate of the altitude data;

(7) checking whether a block capable of splitting remains;

(8) repeating steps (3) to (7) for the block if the dividable block remains as a result of the checking in step (7);

(9) if the block in which division is possible remains as a result of the checking in step (7), terminating division for all blocks; And

(10) constructing an actual cube mesh in three-dimensional space, using the maximum, minimum, and position coordinates of the elevation data stored for each block for which partitioning is complete. Features of the jacket.

Preferably, after setting the end flag for the block in which division is completed in step (6), whether a block capable of dividing in step (7) remains by using whether to set an end flag for each block. You can check it.

Also, preferably, in step (1), the constant value multiplied by the standard deviation may be set to one.

Applicants studied voxel-based terrain rendering techniques, which consist of two stages of preprocessing and rendering for real-time rendering of elevation data. In the preprocessing step, a cube mesh is created that completely covers the surface of the terrain. In the rendering step, the cube mesh generated in the preprocessing step is used to accelerate the voxel-based ray projection method.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram illustrating respective steps of a cube mesh generating method according to an embodiment of the present invention. As shown in FIG. 3, in the method of generating a cube mesh according to the present invention, a step 310 of obtaining altitude data largely, processing 320 information of a quad tree generation step 320, and generating a quad tree step 330. And a step 340 of generating a cube mesh. More specifically, as shown in Figure 3, it consists of 10 steps of S01 ~ S10. That is, in the method of generating a cube mesh according to the present invention, after calculating the standard deviation of the whole altitude data, multiplying the calculated standard deviation by a constant value to calculate a split reference value (S01), and the whole altitude data in one block. Starting the division of the block (S02), calculating the difference between the maximum value and the minimum value of the altitude data for the block (S03), and divides the difference between the calculated maximum and minimum values of the altitude data for the block In step S04, when the difference between the maximum value and the minimum value of the altitude data for the block is greater than the split reference value, comparing the reference value with the reference value (S04), and dividing the block into quarters (S05), and in step S04 As a result of the comparison, if the difference between the maximum value and the minimum value of the altitude data for the corresponding block is equal to or less than the division reference value, the dividing for the corresponding block is terminated (S06), and the maximum value of the altitude data for the corresponding block is finished. Storing the minimum value and the position coordinate (S06_2), checking whether a block capable of dividing is left (S07), and as a result of checking in step S07, if a block capable of dividing remains, steps S03 to S07 for the corresponding block Repeating step (S08), if the result of the check in step S07, if there is no block that can be divided, the step of terminating the division for all blocks (S09), and the stored altitude data for each block is completed Step S10 of generating a real cube mesh in a three-dimensional space by using a maximum value, a minimum value and a position coordinate of.

In particular, after setting the end flag for the block in which the partitioning is completed in step S06, it is possible to check whether there are any blocks that can be divided in step S07 using whether to set the end flag for each block. Further, according to the embodiment, the division reference value may be set equal to the standard deviation by setting the constant value multiplied by the standard deviation to 1 in step S01.

First, a step of dividing the total altitude data using a quad tree will be described in detail. 4 is a diagram illustrating a method of dividing an entire altitude data block using altitude data according to an embodiment of the present invention. As shown in FIG. 4, after calculating the difference between the maximum value and the minimum value of the altitude data with respect to the start block (the entire altitude data block) (250 according to FIG. 4), the calculated difference 250 is divided into a division reference value ( According to FIG. 4, the block is divided into four parts because it is larger than 130 (one step division). After calculating the difference between the maximum value and the minimum value of the altitude data for each block divided into four equal parts (60, 80, 120, 250, respectively, according to FIG. 4), the calculated difference is compared with the split reference value. Determine whether to split the block. According to FIG. 4, since only blocks having a difference of 250 are larger than the division reference value, the blocks are divided into four parts (two step division), and one of the two step division blocks is displayed as four equal divisions (three step division).

Here, the division reference value, which is the basis of the division, is a value obtained by multiplying the standard deviation of the whole altitude data by a constant value, and the constant value becomes a parameter for determining the degree of division. Equations 1 and 2 show equations for calculating the split reference value and the standard deviation, respectively.

)은 표준편차(

)에 분할의 정도를 결정할 수 있는 상수 값(c)을 곱해서 계산될 수 있다. 표준편차(

)는 모든 고도 데이터에 대하여 계산되어야 하는데, 수학식 2의 경우, M x N개의 고도 데이터에 대하여 표준편차가 계산되는 경우이다. 수학식 2에서, avg는 전체 고도 데이터에 대한 평균값을 의미하며, p_{i ,j}는 (i, j)번째 셀의 고도 데이터 값을 나타낸다.In Equation 1, the division reference value (

) Is the standard deviation (

) Can be calculated by multiplying by a constant value (c) which determines the degree of division. Standard Deviation(

) Should be calculated for all altitude data. In Equation 2, the standard deviation is calculated for M x N altitude data. In Equation 2, avg means an average value for the entire altitude data, and p _{i , j} represents the altitude data value of the (i, j) th cell.

Most quad tree-based methods resize partitioned quadrant spaces for speed. This is because the smaller the size of the divided blocks, the shorter the distance between the surrounding polygon and the actual terrain, and the higher the space hopping effect. However, in order to create such a dense cube mesh, many polygons have to be rendered, which results in a decrease in speed. In the present invention, the degree of performance reduction can be reduced by determining the optimal block size through basic parameters such as maximum and minimum values for the entire data in the preprocessing step.

As such, when the partitioning is finished and the maximum, minimum, and position coordinate information is obtained for each block, the next step is to actually generate a cube mesh in the three-dimensional space using the information obtained above. FIG. 5 is a diagram illustrating a process of generating a cube mesh according to an embodiment of the present invention. FIG. 5A is a diagram illustrating maximum and minimum value information for each block obtained when partitioning of blocks is completed. 5B is a diagram illustrating the generation of a cube using the maximum value information of FIG. 5A, and FIG. 5C illustrates a final cube mesh by resizing and reflecting the minimum value information in the cube generated in FIG. 5B and combining them. It is a figure which shows that. In generalizing the process of Figure 5, after generating the unit tube (1 x 1 x 1 size), the generated unit cube is transformed in size using the maximum value and the minimum value information, and the parallel movement to the pre-calculated position Let's do it. After these transformations have been performed for all unit cubes, they can be combined into one to create a perfectly shaped cube mesh with an adaptive quad tree.

6 is a diagram illustrating an example of a cube mesh finally generated according to the present invention. As shown in FIG. 6, since the cube mesh generated according to the present invention completely surrounds the topography of the divided faces, the entire cube mesh in which the respective cubes are combined also penetrates the topography of the surrounding surfaces. Convex problem) does not occur. Since polygon meshes of the conventional envelope method more closely surround the terrain than the cube mesh of the present invention, the speed in the rendering stage is slightly faster than our method. However, our method also makes little difference in speed from the envelope method if we make the cube mesh smaller, i.e. if we split the quad tree a bit more tightly. Furthermore, the cube mesh structure of the present invention may apply the early ray termination method in the rendering step, and thus may render the rendering speed faster than the envelope method. On the other hand, the comparison between the two methods is not only speed, but also the distortion of the resulting image (the problem of penetrating the mesh surrounding the actual terrain) can be a major comparison factor. Next, the performance of the method according to the present invention will be examined through comparison experiments with the envelope method.

The method and envelope method according to the present invention were implemented in a computer with a Pentium IV 3.0 GHz CPU, 1 GB of main memory, and an NVIDIA Geforce 6600 graphics card. Puget altitude data was used in the comparative experiments, which are well known benchmarking data with a resolution of 512 x 512. The size of the viewport is 640 x 480.

 7 is a view comparing the rendering time (time of the rendering step except the preprocessing step) of the rendering method and the envelope method using the cube mesh structure generated according to the present invention. As shown in FIG. 7, the method according to the present invention showed a two times improvement in rendering speed when compared to the envelope method. This is because the cube mesh structure according to the present invention can be applied to the early termination method of the ray can be significantly reduced rendering time.

On the other hand, since the method according to the present invention solves the convex problem of the envelope method, there is a great difference in the quality of the image. 8 is a view comparing the resultant image of the method according to the present invention and the envelope method, Figures 8a and 8b is a view showing the resultant image generated by the envelope method, Figure 8c is a cube mesh according to the present invention A diagram showing a resultant image generated using. 8A and 8B illustrate the case in which block sizes are set to 16 x 16 and 8 x 8, respectively. As can be seen in FIG. 8, the resultant image using the cube mesh generated according to the present invention exhibits much improved quality in the entire image area as compared with the resultant image by the envelope method.

Finally, the cube mesh structure according to the present invention allows an appropriate trade-off between image quality and rendering speed by adjusting the degree of segmentation by adjusting a constant value multiplied by the standard deviation. It is also worth noting.

The present invention described above may be variously modified or applied by those skilled in the art, and the scope of the technical idea according to the present invention should be defined by the following claims.

According to the method of generating a cube mesh structure of the present invention, whether or not to divide a specific block is determined by comparing a difference between the maximum value and the minimum value of the elevation data of the specific block with the split reference value based on the standard deviation of the entire elevation data. By creating the cube mesh structure that surrounds the terrain in three-dimensional space by using the information of the blocks whose segmentation is finally finished, it solves the problems of existing methods for accelerating the voxel-based ray projection method and is effective. Enable terrain rendering.

Claims (3)

(1) obtaining a standard deviation of the whole elevation data using height field data indicating elevation values of the terrain, and then multiplying the obtained standard deviation by a constant value to obtain a split reference value; (2) starting dividing the block using all the altitude data as one block; (3) calculating a difference between the maximum value and the minimum value of the altitude data for the block; (4) comparing the difference between the obtained maximum and minimum values of the altitude data for the corresponding block with the split reference value; (5) dividing the corresponding block into quarters if the difference between the maximum value and the minimum value of the altitude data for the corresponding block is greater than the division reference value as a result of the comparison in the step (4); (6) If the difference between the maximum value and the minimum value of the altitude data for the corresponding block is less than or equal to the division reference value, as a result of the comparison in the step (4), the division of the corresponding block is terminated, and Storing a maximum value, a minimum value, and a position coordinate of the altitude data; (7) checking whether a block capable of splitting remains; (8) repeating steps (3) to (7) for the block if the dividable block remains as a result of the checking in step (7); (9) if the block in which division is possible remains as a result of the checking in step (7), terminating division for all blocks; And (10) generating an actual cube mesh in three-dimensional space using the maximum, minimum, and position coordinates of the altitude data stored for each block after the partitioning is completed; How to generate a cube mesh structure for terrain rendering including a. The method of claim 1, And setting an end flag for the block in which division is completed in the step (6), and then checking whether there are remaining blocks that can be divided in the step (7) by using an end flag set for each block. The method according to claim 1 or 2, Setting the constant value to be multiplied by the standard deviation in step (1).

Images