The Algorithm Research Of Volume Rendering And System Realization Of Meteorology Visualization

Visualization in scientific computing is an important assistant tool of scientific research,it helps scientist and personnel of scientific research to observe and understand the essence and meaning of scientific data. In the later eightieth,the notion of visualization was established,during the ten years,the computer and computing technology developed and the observation facility advanced,we have got abundant data,and it is hundreds of times or thousands of times than the past. Scientists often have to treat data measured by hundreds of megabyte or thousands of megabyte,and have difficulty in treating so large quantity of data. Visualization can help dig the pivotal information of data and become more important in scientific research,so is boosted by application requirement.An important application field of visualization is meteorology visualization. In the field a lot of data is produced,and the data must be visualized,converted to denotation of graphic or image. In order to complete meteorology visualization data management is a key problem. This thesis analyses problems existing in the current data management for visualization system front-end,and gives a new way for data management. In order to prove that it is a feasible way to manage visualization data,we realize a data management system (DMS) for MM5 weather model output data.Algorithm research is main part of visualization research,the thesis refers to two kinds of algorithms for volume rendering:Fourier volume rendering and shear-warp volume rendering. The thesis investigates Fourier volume rendering profoundly and implements a rendering engine,we find some problems of Fourier volume rendering and propose a method to solve the problems,implement an algorithm using projection of macro voxels,and we also propose a new kind of parallel Fourier volume rendering algorithm.In the thesis a volume rendering algorithm is put forward using shear-warp factorization of the viewing transformation and two-pass classification,it utilizes the coherence of volume data,classifies the data for the first pass and encodes the volume data using a run encoding,the encoded data is classified for the second pass when it is rendered. Using the feature for two pass classification,it uses octree data structure speed up the rendering,realizes a very good volume rendering algorithm. It is compared with the other same kind using shear-warp factorization of viewing transformation,40-70 percent more quickly.A lot of problems need to be solved in practical application fields. Within the fields of earth,space,environment and meteorology science,global data visualization is often a problem. Because the earth is a non-outspread sphere,plane vector visualization will induce some errors. We analyze its problems,and propose a vector display method based on cartography and in the sphere coordinate system,evolve the formula. We make experiments and prove the formula is right,new method can get better vector display.