Issei Fujishiro

Optimizing Parallel Performance of Streamline Visualization for Large Distributed Flow Datasets

Parallel performance has been a challenging topic in streamline visualization for large unstructured flow datasets on parallel distributed-memory computers. It depends strongly on domain partitions. Unsuitable partitions often lead to severe load imbalance and high frequent communications among the domain partitions. To address the problem, we present an approach to flow data partitioning taking account of flow directions and features. Multilevel spectral graph bisection method is employed to reduce communication and synchronization overhead among distributed domains. Edge weights in the corresponding adjacent matrix is defined based on an anisotropic local diffusion operator which assigns strong coupling along flow direction and weak coupling orthogonal to flow. Meanwhile, the distributions of seed points and flow features such as vortex structure are also considered in partitioning so as to obtain good load balance. The experimental results are given to show the feasibility and effectiveness of our method.

Interval volume decomposer: a topological approach to volume traversal

The Interval Volume Decomposer (IVD) is an interface for decomposing an entire volume into interval volumes each of which characterizes a distinctive volume feature. The advantage of the IVD is that it allows us to look inside the volume by peeling interval volumes from outside to inside not only interactively but also automatically. This is achieved due to the rigorous analysis of nested structures of the decomposed interval volumes by constructing a level-set graph that delineates isosurface transitions according to the scalar field. A robust algorithm for computing such level-set graphs is introduced in order to extract significant structures in the volume by putting together local interval volumes into a finite number of global groups. Several decomposition examples of medical and simulated datasets are demonstrated so that the present interface effectively traverses the underlying structures of the volume.

T-map: a topological approach to visual exploration of time-varying volume data

The rapid advance in high performance computing and measurement technologies has recently made it possible to produce a stupendous amount of time-varying volume datasets in a variety of disciplines. However, there exist a few known visual exploration tools that allow us to investigate the core of their complex dynamics effectively. In this paper, our previous approach to topological volume skeletonization is extended to capture the topological features of large-scale time-varying volume datasets. A visual exploration tool, termed T-map, is presented, where pixel-oriented information visualization techniques are deployed so that the user can identify partial 4D spatiotemporal domains with characteristic changes in a topological sense, prior to detailed and comprehensible volume visualization. A case study with datasets from atomic collision research is performed to illustrate the feasibility of the proposed tool.

Optimization Strategies Using Hybrid MPI+OpenMP Parallelization for Large-Scale Data Visualization on Earth Simulator

An efficient parallel visualization library has been developed for the Earth Simulator. Due to its SMP cluster architecture, a three-level hybrid parallel programming model, including message passing for inter-SMP node communication, loop directives by OpenMP for intra-SMP node parallelization and vectorization for each processing element (PE) was adopted. In order to get good speedup performance with OpenMP parallelization, many strategies are used in implementing the visualization modules such as thread parallelization by OpenMP considering seed point distributions and flow features for parallel streamline generation, multi-coloring reordering to avoid data race of shared variables, some kinds of coherence removal, and hybrid image-space and object-space parallel for volume rendering. Experimental results on the Earth Simulator demonstrate the feasibility and effectiveness of our methods.

Designing 6DOF haptic transfer functions for effective exploration of 3D diffusion tensor fields

Rapid dissemination of DT-MRI devices has recently evoked intensive research to come up with effective methods for grasping the structural features of acquired 3D diffusion tensor fields. In this paper, we further explore the potential of haptization in conjunction with visualization to represent different levels of information on the DT-MRI datasets. The main contribution of this paper is to propose a sophisticated method for utilizing diffusion-based tractography textures mapped on cross-sections to visualize local structures such as nerve tracts, together with appropriate 6DOF haptic transfer functions to haptize elementary diffusion properties. We illustrate the effectiveness of a prototype system featuring a PHANToM Premium 3.0 with an application to a practical DT-MRI human brain dataset.

Feature-Driven Volume Fairing

Volume datasets have been a primary representation for scientific visualization with the advent of rendering algorithms such as marching cubes and ray casting. Nonetheless, illuminating the underlying spatial structures still requires careful adjustment of visualization parameters each time when a different dataset is provided. This paper introduces a new framework, called feature-driven volume fairing, which transforms any 3D scalar field into a canonical form to be used as communication media of scientific volume data. The transformation is accomplished by first modulating the topological structure of the volume so that the associated isosurfaces never incur internal voids, and then geometrically elongating the significant feature regions over the range of scalar field values. This framework allows us to elucidate spatial structures in the volume instantly using a predefined set of visualization parameters, and further enables data compression of the volume with a smaller number of quantization levels for efficient data transmission.

Visualization Research is Growing and Expanding

Visualization has become an increasingly active area of research because of its usefulness in a wide range of applications. APVIS is a meeting covering topics mostly in graph drawing and visualization. At APVIS 2007, we discussed expanding the scope of the venue and attracting greater international participation.

Collaborative Visualization: Topological Approaches to Parameter Tweaking for Informative Volume Rendering

Data crisis ubiquitously arises in solving complex scientific problems. It has motivated us to develop a Collaborative Visualization Environment (CVE), which provides the users with the “serendipity” with the aid of effective data-centric tweaking of visualization-related parameters. To this end, we have proposed the concept of Volume Data Mining (VDM), which takes full advantage of knowledge in the field of differential topology to allow the users to explore the global structures and local features of target 4D volumes. In this paper, the effectiveness of our current CVE framework with several VDM tools is illustrated with applications to practical scientific simulation datasets.

Locating an Optimal Light Source for Volume Rendering

Locating an optimal light source is an important task for image synthesis because it influences the spatial perception of the observers. Several methods have been proposed, that search for the optimal location of light sources for 3D surfaces, but none has been done for volumes. This paper presents a new method for finding an optimal location of a single parallel light source for volumetric ray-casting. The method computes the entropy of brightness distribution of representative isosurfaces. The method then utilizes the opacity transfer function to assign a distinct weight to each of the values of isosurface illumination entropy to evaluate the plausible illuminating entropy for the volume. The light with the maximum entropy value allows the observers to realize the volume features with enriched 3D visual cues.

Hierarchical causality explorer: making complemental use of 3D/2D visualizations

Hierarchical causality relationships reside ubiquitously in the reality. Since the relationships take intricate forms with two kinds of links - hierarchical abstraction and causal association, there exists no single visualization style that allows the user to comprehend them effectively. This paper introduces a novel information visualization framework which can change existing 3D and 2D display styles interactively according to the users visual analysis demands. The two visualization styles play a complementary role, and the change in the style relies on morphing so as to maintain the users cognitive map. Based on this framework, we have developed a general-purpose prototype system, which provides the user with an enriched set of functions not only for supporting fundamental information seeking, but bridging analytic gaps to accomplishing high-level analytic tasks such as knowledge discovery and decision making. The effectiveness of the system is illustrated with an application to the analysis of a nuclear-hazard cover-up problem.