Efficient Local Statistical Analysis via Point-Wise Histograms in Tetrahedral Meshes and Curvilinear Grids

Bo Zhou, Yi-Jen Chiang, Cong Wang

    Research output: Contribution to journalArticle

    Abstract

    histograms (i.e., point-wise histograms computed from local regions of mesh vertices) have been used in many data analysis and visualization applications. Previous methods for computing local histograms are mainly for regular or rectilinear grids. In this paper, we develop theory and novel algorithms for computing local histograms in tetrahedral meshes and curvilinear grids. Our algorithms are theoretically sound and efficient, and work effectively and fast in practice. Our main focus is on scalar fields, but the algorithms also work for vector fields as a by-product with small, easy modifications. Our methods can benefit information theoretic and other distribution-driven analysis. The experiments demonstrate the efficacy of our new techniques, including a utility case study on tetrahedral vector field visualization.

    Original languageEnglish (US)
    JournalIEEE Transactions on Visualization and Computer Graphics
    DOIs
    StateAccepted/In press - Jan 22 2018

      Fingerprint

    Keywords

    • Algorithm design and analysis
    • Entropy
    • Geometry-Based Techniques
    • Histograms
    • Isosurfaces
    • Mathematical Foundations for Visualization
    • Scalar Field Data
    • Statistical analysis
    • Tetrahedral Meshes and Curvilinear Grids
    • Uncertainty
    • Vector Field Data

    ASJC Scopus subject areas

    • Software
    • Signal Processing
    • Computer Vision and Pattern Recognition
    • Computer Graphics and Computer-Aided Design

    Cite this