VoxelFlow: A Parallel Volume Rendering Method for Scientific Visualization

Volume Visualization techniques are often applied to sampled data obtained from MRI and CT-scans as well as to simulated data computed by computational fluid dynamics (CFD) methods. In such applications, a sequence of images is animated by incrementally rotating the volumetric object about its center, so that the scientist can view the structure from all angles. In this paper, we propose an efficient method, called VoxelFlow, for generating rendered images of volume data which can be used in animating rotations of objects. The algorithm uses a look-ahead scheme of voxel dataflow and exploits the communication features of modern scalable multicomputers to achieve near linear speedup. Simulated results are shown for implementations on mesh architectures like the Intel Paragon and torus architectures like the Cray T3D. Computational and communicational load balancing issues among processors are also addressed. 1. Introduction Medical data obtained from MRI and CT-scanners act as good sou..