Research On Feature-Based Visualization For Time-Varying Flow Volume

Visualization is a technique which was original utilized in scientific computing field. Creating images, diagrams, or animations to communicate a message is its basic purpose. Recently, Visualization has ever-increasing applications in Computer Aid Design (CAD), medical analysis, information processing, etc.The main content of this thesis is feature-based visualization while the objects are the ?uids obtained from hydromechanics and aerodynamics. Based on the feature analysis process, our final goal is to develop a way to depict the evolution of time-varying features with a minimum actual cost, and assist the scientists to create a visual indispensable tool for extracting knowledge from large amounts of data which could also be utilized to improve their experiments and simulations.A brief overview of the origin, development and the state of the art of the visualization technique have been presented firstly in the thesis, and the time-varying volume data has been introduced based on the discussion of the structure of the grids. Then, the analysis and comparison of feature-based ?uid visualization have been given, also a opacity threshold based feature extraction, a prediction-correction based feature tracking and ?uid feature event detection methods have been proposed, so the feature could be selected and separated from a complex environment. Meanwhile, in order to depict the evolution of ?uid features by utilizing a picture, the post-processing step of the time-varying feature has been mentioned, including feature storing, management and the fade-in rendering. Finally, an interactive similarity measurement based feature filtering, comparison, extraction and tracking approach has been discussed which adopts several user-defined measurements to obtain the similarities between di?erent ?uid features.Our experiment results show the comprehension of the entire volume could be derived through our method with the advantage of e?cient spatial and temporal consumption. By considering the similarity measurement between di?erent features, detail time-varying information of the volume could be conveyed by our approach which creates a foundation for users to have a better understanding of the data.