Date of Award
2015
Document Type
Open Access Dissertation
Degree Name
Doctor of Philosophy in Computer Science (PhD)
Administrative Home Department
Department of Computer Science
Advisor 1
Ching-Kuang Shene
Advisor 2
Chaoli Wang
Committee Member 1
Laura E. Brown
Committee Member 2
Raymond A. Shaw
Abstract
Flow visualization plays an important role in many scientific and engineering disciplines such as climate modeling, turbulent combustion, and automobile design. The most common method for flow visualization is to display integral flow lines such as streamlines computed from particle tracing. Effective streamline visualization should capture flow patterns and display them with appropriate density, so that critical flow information can be visually acquired. In this dissertation, we present several approaches that facilitate expressive flow field visualization and exploration. First, we design a unified information-theoretic framework to model streamline selection and viewpoint selection as symmetric problems. Two interrelated information channels are constructed between a pool of candidate streamlines and a set of sample viewpoints. Based on these information channels, we define streamline information and viewpoint information to select best streamlines and viewpoints, respectively. Second, we present a focus+context framework to magnify small features and reduce occlusion around them while compacting the context region in a full view. This framework parititions the volume into blocks and deforms them to guide streamline repositioning. The desired deformation is formulated into energy terms and achieved by minimizing the energy function. Third, measuring the similarity of integral curves is fundamental to many tasks such as feature detection, pattern querying, streamline clustering and hierarchical exploration. We introduce FlowString that extracts shape invariant features from streamlines to form an alphabet of characters, and encodes each streamline into a string. The similarity of two streamline segments then becomes a specially designed edit distance between two strings. Leveraging the suffix tree, FlowString provides a string-based method for exploratory streamline analysis and visualization. A universal alphabet is learned from multiple data sets to capture basic flow patterns that exist in a variety of flow fields. This allows easy comparison and efficient query across data sets. Fourth, for exploration of vascular data sets, which contain a series of vector fields together with multiple scalar fields, we design a web-based approach for users to investigate the relationship among different properties guided by histograms. The vessel structure is mapped from the 3D volume space to a 2D graph, which allow more efficient interaction and effective visualization on websites. A segmentation scheme is proposed to divide the vessel structure based on a user specified property to further explore the distribution of that property over space.
Recommended Citation
Tao, Jun, "ENABLING TECHNIQUES FOR EXPRESSIVE FLOW FIELD VISUALIZATION AND EXPLORATION", Open Access Dissertation, Michigan Technological University, 2015.