Visualizing and manipulating volumetric data is a major component in many areas including anatomical registration in biomedical fields, seismic data analysis in the oil industry, machine part design in computer-aided geometric design, character animation in the movie industry, and fluid simulation. These industries have to meet the demands of the times and be able to make meaningful assertions about the data they generate. The shear size of this data presents many challenges to facilitating realtime interaction. In the recent decade, graphics hardware has become increasingly powerful and more sophisticated which has introduced a new realm of possibilities for processing volumetric data. This thesis focuses on a suite of techniques for viewing and editing volumetric data that efficiently use the processing power of central processing units (CPUs) as well as the large processing power of the graphics hardware (GPUs). This work begins with an algorithm to improve the efficiency of a texture-based volume rendering. We continue with a framework for performing realtime constructive solid geometry (CSG) with complex shapes and smoothing operations on watertight meshes based on a variation of Depth Peeling. We then move to an intuitive technique for deforming volumetric data using a collection of control points. Finally, we apply this technique to image registration of 3-dimensional computed tomography (CT) images used for lung cancel treatment, planning.
展开▼
