Research On Isosurface Extraction Of Large Scale Datasets With Out-of-Core Method

Visualization has been considered as an important method for analyzing and processing large-scale dataset. To processing large-dataset, people must overcome lots of difficulties, such as preprocessing time is too long, rendering time is too long or consuming too much space. How to realize the real-time and interactive visualization of large scale dataset has become a hot topic of scientific visualization.We will discuss the pre-processing methods, parallel rendering.technology and accelerateing render technology in this paper. For solving the problems of large-scale dataset visualization, we will combine many techonologies such as Coordinates Deviding, Scalar Deviding, Out-of-Core Technology and Parallel Render and so on.Then we will put forward BBIO tree method based on Coordinates Deviding and Compress strategy based on Span Space Deviding technology. At last, a Large-scale dateset Visualization Frame will be realized integrateing two method we put forwarded. The contributions and relevant work in the paper are as follows.Firstly, a Four End-Point algorithm is presented for accelerateing searching BBIO tree node. The traditional BBIO tree constructing method usually builds ill tree with bad struct. The Four End-Point algorithm offset the effect of bad struct, and it is a method based on re-sortingof interval. The method has reduced the time of searching tree under 0.001.second.Secondly, BBIO Tree Based on Two-Dimensionality Coordinates Deviding is presented to realize large-scale dataset visualization. Through deviding dataset by two-dimensionality, visualzition is fit to display with Tiled-Display Wall. In phase of rendering, Interval Mark Method is used to reduce isosurface extraction time 23%.Thirdly, Compress Index Method Based on Span Space is put forward to sovle the problem of which is hard to balance workload in visualization processing. The large-scale dataset devided by Span Space, and through parallel merging and all-to-all communicating, the workload will be assured balance. A Compress Index Method is presented to compress the tree struct. It will reduce the index struct space about 40%. Forthly, Binary Interval Tree Based on Sub-Meta-Cell is presented in phrase of rendering. It is a method which relateless with the iso-surface, and the render time will reduce 30% compared with traditional Marching Cubes method.Finally, the paper realized of all method discussing above.Followed by the goals and the principia of modern Software Engineering, we build a large-scale dataset visualizing frame, and it is a scalable and easy-reused frame struct.