Integral surface based flow field visualization

The focus of this research is in the area of vector field and flow visualization. Such a study is important in order to simplify the raw data of complex vector fields and make them more understandable to scientists. Challenges faced by flow filed visualization methods include performance, perception and large data size. Two main aspects of vector field visualization have been studied in this research to solve different issues belong to these categories, including visualization method and flow field abstraction. For visualization, two GPU accelerated time surface methods have been introduced. The first method, which is developed to visualize strange attractors, is based on adaptive subdivision of a polygonal object deforming under the vector field. Simplification is also applied to the polygonal object to limit the complexity, alleviate self-intersection issues and handle high curvature features. The second method uses a point-based time surface to resolve the difficulty of mesh decimation in the first method and takes advantage of the parallel architecture of the modern GPU. This method has been applied to general flow field data sets and can provide visually identical result to the geometry-based approach. In addition to these visualization methods, we have also developed an illustrative visualization framework which generates a concise global visualization using a small number of streamribbon like visual metaphors, named streamtapes. This method starts with a set of initial streamlines. Each of them is given a simple descriptor. A two level clustering method is applied to the descriptors to group similar streamlines. Representative streamtapes are generated for each cluster to provide the final result. We demonstrate the usefulness of this method with large and complex synthetic and real data sets, and show that our visualization results can convey the structure information of the flow field data in a concise and informative way.