Abstract
Applications in scientific visualization often involve extensive data sets, which represent scalar or vector valued functions resulting from experimental measurements. There are two major approaches for visualizing a set of multi-dimensional volume data, namely, the direct volume rendering (DVR) and the isosurface method. However, both approaches involve manipulating a large size of data files and usually require extensive execution time, which may disrupt interactive applications. In this study, we propose algorithms to cope with the storing, transmitting, and rendering problems in visualizing volume data. First, we study the compression problem of Marching Cubes (MC) isosurfaces, and introduce a geometry compression algorithm to reduce the number of bits of generated triangle data. The introduced algorithm encodes each triangle-vertex according to the index of its containing cube, the index of its containing cube-edge, and its relative position on the cube-edge. Furthermore, the connectivity among triangle-vertices in a cube is encoded by the signs of its vertices, computed by comparing their values to the isosurface threshold. Both theoretical analysis and experimental results show that the introduced algorithm can achieve an excellent compression ratio. Then, we extend the geometry compression algorithm to an octree-based adaptive MC algorithm such that both the number of triangles and the number of bits required to represent a triangle can be reduced simultaneously. Finally, we propose a novel approach to improve a ray-casting based DVR algorithm. The basic idea of our approach is to incorporate the notion of coherence in the resampling process so that the time consuming invocations of trigonometric functions and matrix multiplication calculations can be reduced significantly.