完整後設資料紀錄
DC 欄位語言
dc.contributor.authorFahn, Chin-Shyurng
dc.contributor.authorWu, Shih-Tung
dc.date.accessioned2009-06-02T07:20:40Z
dc.date.accessioned2020-05-29T06:18:20Z-
dc.date.available2009-06-02T07:20:40Z
dc.date.available2020-05-29T06:18:20Z-
dc.date.issued2006-11-13T01:38:58Z
dc.date.submitted1999-12-20
dc.identifier.urihttp://dspace.fcu.edu.tw/handle/2377/3127-
dc.description.abstractThis paper presents a straightforward and efficient algorithm for the view-dependent real-time rendering of large-scale terrain in continuous levels of detail (LOD). It can render the terrain in continuous LOD without the overhead of tree structures or off-line computations of supplemental data. Such a terrain simplification algorithm consists of two main processes that we call block-based concatenation and vertex-based refinement for each visible block. Before the block-based concatenation process, a piece of terrain had been partitioned into small units referred to as blocks. In the course of matching adjacent blocks, the LOD value for each block is first determined according to the importance of local geometry and the distance from an observer to the foreground formed by the terrain. The difference between the LOD of adjacent blocks is set to be at most one. To achieve this, a look-up table is employed to record all the cases of boundary matching. In this manner, T-junctions or cracks occurring in the boundaries of blocks can be avoided with no additional effort. In the vertex-based refinement process, a screen-space error metric is used as a vertex refinement criterion. And the estimation of the error is on a pixel level. Prior to vertex refinement, each block is further divided into a set of sub-blocks; after that, they are manipulated from the outside to the center within a block in clockwise spiral order. During the vertex refinement, all the non-boundary vertices of the currently processed block are measured by an error function. If the error associated with a vertex is smaller than a specified pixel error, the vertex is considered to remove. Conversely, the vertex is reserved. The experimental terrain data are originally generated by a fractal algorithm, and followed by manual editing. A primary terrain rendering system that we have developed is demonstrated on general PC platforms. In this system, the terrain simplification algorithm has been implemented for approximating and rendering digital terrain models as well as the other kinds of height fields, which consistently performs at interactive frame rates of high image quality.
dc.description.sponsorship淡江大學, 台北縣
dc.format.extent8p.
dc.format.extent1341029 bytes
dc.format.mimetypeapplication/pdf
dc.language.isozh_TW
dc.relation.ispartofseries1999 NCS會議
dc.subjectLevels of detail
dc.subjectterrain rendering
dc.subjectgeometry reduction
dc.subjectview-frustum culling
dc.subjectgeomorphing
dc.subject.otherVirtual Reality
dc.titleThe View-Dependent Real-Time Rendering of Large-Scale Terrain in Continuous Levels of Detail
分類:1999年 NCS 全國計算機會議

文件中的檔案:
檔案 描述 大小格式 
ce07ncs001999000129.pdf1.32 MBAdobe PDF檢視/開啟


在 DSpace 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。