| Scientific data visualization is a new and promising field in which much application-oriented research has already been done. A uniform and consistent description of the fundamentals of scientific data visualization, however, is missing. In this dissertation an application-independent framework for scientific data visualization, in which surface-based visualization methods are emphasized, is proposed. This framework is based on a hierarchical data classification, and is used to analyze and describe the entire visualization process, from original data set to the final image. Based on this framework, a Bezier-based visualization model for scalar data is proposed. This model, like any other, requires more input data than is usually available. This has led to the development of a triangular Bezier interpolant that uses six parameters instead of the original nine. An application of this method to actual biomedical data is shown. In addition, a new and faster algorithm for subdivision of tensor-product Bezier surfaces is developed, which is important because this algorithm is repeatedly used for displaying a Bezier surface on a computer screen. Tensor-product Bezier surfaces may be used for interpolation of regular gridded surface data, which is one of the most often encountered data classes within scientific data visualization. |
