Relation-Aware Volume Exploration Pipeline

Ming-Yuen Chan²     Huamin Qu²     Ka-Kei Chung²     Wai-Ho Mak²     Yingcai Wu^1
1Microsoft Research Asia      ²Hong Kong University of Science and Technology

Exploration on a human body dataset: (a) An object (bullet) was inserted into the human body; (b) A relation graph was constructed and(c) the nodes were clustered according to their class to reduce visual clutter; (d) Nodes related to the object were extracted by relation queries; (e) The extracted nodes were expanded for detailed analysis; (f) Images were generated for revealing the selected relations.

Volume exploration is an important issue in scientific visualization. Research on volume exploration has been focused on revealing hidden structures in volumetric data. While the information of individual structures or features is useful in practice, spatial relations between structures are also important in many applications and can provide further insights into the data. In this paper, we systematically study the extraction, representation, exploration, and visualization of spatial relations in volumetric data and propose a novel relation-aware visualization pipeline for volume exploration. In our pipeline, various relations in the volume are first defined and measured using region connection calculus (RCC) and then represented using a graph interface called relation graph. With RCC and the relation graph, relation query and interactive exploration can be conducted in a comprehensive and intuitive way. The visualization process is further assisted with relation-revealing viewpoint selection and color and opacity enhancement. We also introduce a quality assessment scheme which evaluates the perception of spatial relations in the rendered images. Experiments on various datasets demonstrate the practical use of our system in exploratory visualization.

@article {YWu2009b,
author = {Ming-Yuen Chan and Huamin Qu and Ka-Lei Chung and Wai-Ho Mak and Yingcai Wu},
title = {Relation-Aware Volume Exploration Pipeline} ,
journal = {IEEE Transactions on Visualization and Computer Graphics},
year = {2008},
volume = {14},
number = {6},
pages = {1683--1690 }
}