KR20020091733A - Multiresolution Terrain Model Based on Binary Segmentation - Google Patents

Abstract

PURPOSE: A multi-resolution topography model based on a binary division method is provided to reduce costs with respect to a storage space, generate topography rapidly and satisfy a quality level of required topography data. CONSTITUTION: Topography is spatially divided by binary division. Each of divided triangles receives a unique index and stores an error value for extracting a fixed resolution. The error value corresponds to a maximum distance between planes constructing the triangle and topography data included in the planes. A topography generation procedure is repeated until all topography data constructs a triangle. A phase relation between the divided triangles has a cluster form.

Description

Multi-resolution Terrain Model Based on Binary Segmentation

The present invention relates to a multi-resolution terrain model that can process a large amount of terrain information in real time, and more particularly, to define a phase relationship between triangles generated using binary segmentation and to meet various user requirements. The present invention relates to algorithms for modeling terrains for storage costs and real-time processing.

Terrain data is a continuous function space that assigns altitude values to all points of a two-dimensional plane, and all points of a two-dimensional plane are mapped to only one altitude value, so they are called 2.5-dimensional objects to distinguish them from other three-dimensional objects. It is used as important basic information in many different areas such as GIS, virtual environment simulation, finite element analysis, computer vision and computer graphics.

In general, in a visual application field, a single piece of terrain data can be used for different purposes, and the quality of the terrain data required varies depending on the purpose of use, so that a single terrain model can be developed to satisfy various quality levels. This is essential, and as a result, various types of multi-resolution terrain models have been developed. Here, the multi-resolution terrain model is a terrain model made to extract and use terrain data having a certain range of resolution from one terrain representation, and the resolution of the terrain refers to the degree of distinguishing each feature of the terrain.

Multi-resolution terrain models are largely divided into layered and unified models. The stacked model has a conceptual framework that hierarchically stores terrain data satisfying several levels of resolution in a single data structure, and terrain data at a resolution level not stored in the terrain representation is already stored. Extract through a combination. However, there is a spatial discontinuity between different topographical data, and in solving this problem, the resolution level drops below the required level and the amount of data also increases. As an alternative to this problem, an integration model has been proposed. The integrated model is a terrain model designed to express a range of resolution levels freely, and approaches such as history management, transformation, and recursive subdivision are mainly used. Here, the history management-based integrated model uses an approach that obtains the terrain representation by structurally storing the results of the terrain approximation process according to time and spatial relationship. However, the process of creating the terrain is complicated and the storage space is expensive. The integrated model using the transformation technique uses the transformation result obtained through the transformation function of Wavelett as a terrain representation, and extracts an appropriate level of terrain data by appropriately adjusting the transformation coefficients during the inverse transformation process. While the performance is very fast, the spatial relationship between the various resolution terrain meshes is unclear, and the main terrain elements included in the input terrain are blunted as the resolution changes.

Finally, the recursive partitioning model is a hierarchical terrain model obtained by repeating the process of dividing the entire space regularly and repeatedly. The first study of the multi-resolution terrain model based on the hierarchical method was based on QuadTree as shown in FIG. The top node of the QuadTree represents the entire terrain, and the terrain data is repeatedly divided into four quadrangles as shown in (a) of FIG. 1 until the terrain distortion of each leaf node is less than the allowable level. Very similar to the QuadTree model, there is a Quaternary triangulation model as shown in FIG. The only difference is that the division of the terrain is done in triangular units rather than rectangular units as shown in FIG. Both QuadTree-based models and Quaternary triangulation models have the disadvantage that spatial inconsistency occurs on the partition boundary, and the Ternary triangulation model proposed by De Floriani has the largest error in the triangular patch as shown in Fig. 1 (c). We solved the problem of spatial inconsistency by dividing the domain by connecting with the other three vertices. However, the Ternary triangulation model has a disadvantage in that the shape of the generated triangular patch is excessively distorted.

As described above, the present invention eliminates spatial inconsistency in terms of terrain segmentation economy and at the same time reduces the cost of storage space, and provides a terrain model capable of satisfying the quality level of terrain data required for fast terrain generation and use purpose. Its purpose is to develop it.

In order to achieve this object, the present invention provides a data structure for spatially partitioning a large amount of terrain information based on a binary partitioning method, a topological relationship between partitioned triangles, a terrain generation algorithm for generating indexes of triangles from terrain data, and fixing. The fixed resolution terrain model extraction algorithm based on the error value, and the variable resolution terrain model extraction algorithm using the area of the triangle projected on the screen as an error value.

1 is a process in which the terrain is spatially divided by binary division

2 is a phase relationship between the divided triangles

3 is an algorithm for finding three vertices of triangle using the relationship of FIG.

4 illustrates a case in which spatial discontinuity may occur between triangles when terrain data is divided into triangles due to an error.

5 is a fixed resolution terrain extraction algorithm

6 is a view of a terrain model generated by a fixed resolution terrain extraction algorithm and a terrain generated by a variable resolution terrain extraction algorithm;

Hereinafter, described in detail by the accompanying drawings as follows.

Figure 1 (a) shows the shape that the terrain is spatially divided by the binary division method. Each divided triangle is assigned a unique index and stores an error value for fixed resolution extraction. This error value is the maximum distance between the triangular plane and the terrain data contained within it. The terrain generation process continues as shown in FIG. 1 until all terrain data form a triangle. Figure 1 (b) shows the hat relationship between the triangles when the terrain space is divided. The phase relationship between the divided triangles has a cluster shape as shown in FIG. 2. The cluster shows the shape of a triangle that can be represented by binary division, and is divided into two cases. 2 (a) is called an odd cluster and FIG. 2 (c) is called an even cluster, and each divided triangle must have one of (a) or (c) of FIG. 2 to obtain three vertices of the triangle. It is used to The cluster consists of its own length, center point and phase. The partitioned triangular index is obtained by a simple binary operation to obtain its length, center point, and phase. 2 (b) and (d) of FIG. 2 show the phase change relationship of the divided triangle when each triangle is divided. (D) and (E) of FIG. 2 show relational expressions for finding three vertices with a divided triangle using a current phase, a length, and a center point, and FIG. 3 shows an algorithm for finding three vertices of a triangle.

Using the above relationship, the fixed resolution terrain extraction algorithm extracts the terrain by the user's error so as to have a desired level of resolution in the terrain already created as shown in FIG. 4 shows that spatial discontinuity may occur between triangles when terrain data is divided into triangles due to an error. It is possible to solve the spatial discontinuity problem by forcibly dividing adjacent triangles, and the adjacent triangles can be easily identified by the phase relationship of triangle indices. Triangles not visible at the viewer's location and triangles outside the viewing volume are excluded for quick extraction of the terrain. The fixed resolution terrain extraction algorithm is shown in FIG. The variable resolution terrain extraction algorithm is not very different from the fixed resolution terrain extraction algorithm. Only the area of the projected triangles on the screen is used as a user error value to determine whether to divide the terrain.

The algorithms described above can reduce the number of triangles while approximating the terrain to the actual terrain by user error. 6A is a terrain model generated by a fixed resolution terrain extraction algorithm. 6B shows the appearance of the terrain generated by the variable resolution terrain extraction algorithm. Here, it is possible to show the ability to process the terrain close to the actual terrain in real time without causing severe distortion of the terrain by appropriate user error.

As described above, the present invention provides an advantage of processing a large amount of terrain information in real time and various levels of resolution. Spatial partitioning of the terrain by binary partitioning makes the phase relationship between triangles clear, which makes it possible to reduce the cost of storage space and to apply it to the desktop environment. In the future, it is expected to be widely used in terrain information processing fields such as 3D GIS and terrain-based visual simulation.

Claims (1)

Regarding multi-resolution terrain models that can process large amounts of terrain information in real time, Find the three vertices and the phase relationship between the divided triangles based on binary division Terrain generation algorithm, Fixed resolution terrain algorithm, Variable resolution terrain algorithm.