A Vocabulary Approach to Partial Streamline Matching and Exploratory Flow Visualization

Area of Science:

• Computational fluid dynamics
• Data visualization
• Scientific computing

Background:

• Streamline similarity measurement is crucial for flow data analysis tasks like feature detection and clustering.
• Existing methods often struggle with shape invariance under transformations such as translation, rotation, and scaling.

Purpose of the Study:

• To introduce FlowString, a novel vocabulary-based approach for streamline similarity analysis and visualization.
• To develop a method that captures intrinsic streamline similarity invariant to geometric transformations.
• To enable flexible and intuitive querying for streamline data.

Main Methods:

• Resampling streamlines based on local feature scales.
• Classifying points along streamlines using local shape similarity.
• Encoding streamlines into shape-character strings for querying and retrieval.
• Developing an interface for exact and approximate partial streamline matching.

Main Results:

• FlowString successfully extracts shape-invariant features, enabling robust streamline similarity measurement.
• The string-based encoding allows for efficient querying and retrieval of streamline patterns.
• Demonstrated effectiveness across diverse flow field datasets.
• Extended functionality for multi-dataset analysis.

Conclusions:

• FlowString provides a powerful and intuitive method for exploratory streamline analysis and visualization.
• The approach offers intrinsic similarity measurement invariant to common transformations.
• User evaluations confirm the approach's usefulness for flow data exploration and pattern discovery.