Feature driven compression and simplification of two-dimensional vector fields

Compressing and simplifying dense vector fields is an important task that has only recently been addressed. The end result of vector field compression techniques is a reduced dataset that has potentially lost its descriptive features. Analyzing these compressed datasets can lead to incorrect interpretation of flow. Researchers also argue in favor of simplifying features, since standard vector field visualization techniques can result in cluttered images that are difficult to interpret. Our research is concerned with the study of vector field feature driven compression and simplification.; In the area of feature driven compression we present one bottom-up and two top-down algorithms for compressing 2D vector fields that preserve topology. Our approach is to reduce the size of a given dataset with constrained compression techniques that considers topological information. We employ different types of global metrics to measure the degradation in topology and incorporate some local error metrics to evaluate the change in the underlying dataset. Results indicate that our compression techniques can significantly reduce the dataset and preserve topology while maintaining low local errors.; In the area of feature driven simplification we introduce a topology simplification algorithm that can significantly reduce topological clutter while maintaining the global structure of the original topology. Our approach is flexible, as it can be used with any critical point importance metric or a combination of metrics. Simplification is driven by a constrained simplification technique that incorporates critical point importance. To evaluate the change in the underlying dataset we utilize local angular and magnitude errors. Our simplification algorithm can also be used as pre-processing step for our bottom-up compression scheme. Results indicate that one can obtain significant simplification with low errors without losing important topological information.