Multiscale Symmetry Detection in Scalar Fields by Clustering Contours

Area of Science:

• Scientific Visualization
• Data Analysis
• Computational Geometry

Background:

• Visualizing complex volumetric data, particularly scalar fields, presents significant challenges for direct exploration.
• Isocontour extraction is a common technique for simplifying scalar field visualization by highlighting data features.
• Detecting symmetry in scalar fields is difficult, requiring both data segmentation and identification of invariant features.

Purpose of the Study:

• To develop a novel representation for contours to study their similarity.
• To design a clustering-based algorithm for detecting symmetric regions in scalar fields.
• To combine topological analysis and geometric solutions for robust symmetry detection.

Main Methods:

• Contours are mapped to a high-dimensional, transformation-invariant descriptor space.
• A clustering algorithm is employed to identify similar contour representations.
• The contour tree is utilized for data segmentation, and the descriptor space for transformation invariance.

Main Results:

• A novel contour representation facilitates the study of similarity relationships between contours.
• The proposed clustering-based algorithm effectively detects symmetric regions within scalar fields.
• The integration of contour trees and descriptor spaces provides a robust approach to symmetry detection.

Conclusions:

• The developed method offers a powerful tool for exploring and understanding complex scalar fields.
• Applications in query-driven exploration and asymmetry visualization demonstrate the approach's effectiveness.
• This work advances the field of scientific visualization by enabling more sophisticated analysis of scalar field data.