The real time processing of very large volumes introduces specific algorithmic challenges due to the impossibility of fitting the input data in the main memory of a computer. The basic assumption (RAM computational model) of uniform-constant-time access to each memory location is not valid and the performance of most algorithms does not scale due to the high frequency of I/O operations. In this talk I will present recent research results in out-of-core computing addressing specifically the issues of algorithm redesign and data layout restructuring that enable data access patterns with minimal performance degradation in external memory computations. The effectiveness of the approach is demonstrated with a prototype visualization tool allowing interactive exploration of large scalar fields using very modest computing resources. Unprecedented results are obtained both in terms of absolute performance and, more importantly, in terms of scalability. On a laptop computer we provide real time interaction with a 2048^3 grid (8 Giga-nodes) using only 20MB of memory. The scheme relies simply on the determination of a cache oblivious reordering of the rectilinear grid data and a progressive construction of the rendering output. The cache oblivious reordering minimizes the amount of I/O performed during the out-of-core computation. The progressive and asynchronous computation of the output provides flexible quality-for-speed tradeoffs within a time-critical interruptible user interface. A data-streaming infrastructure provides immediate network access to large datasets stored on remote data servers. I will conclude the talk with a live demonstration with the current prototype implementation of the scheme.

The importance of exploratory data analysis (EDA) as a prerequisite to application of computational methods, such as traditional statistical analysis, is currently widely recognized. The goal of EDA is to gain understanding of data, i.e. to penetrate into relationships, patterns, and trends hidden inside data and to formulate hypotheses that can later be checked using statistical methods. Preliminary investigation of data must also precede their preparation to processing by various computation-based analysis tools, such as data mining. Techniques of EDA are mostly based on data visualization, i.e. the graphical presentation of data in ways that prompt the discovery of important traits and relationships. Computers enabled features of graphical presentations that are now considered indispensable for EDA: high user interactivity, allowance for various transformations, and multiple dynamically linked views such that changes in one display are immediately propagated to all others. An important category of data dealt with in statistics is spatially referenced data. For visualization of such data, maps are traditionally used, since they are isomorphic to space and thus capable of representing and conveying to human's eye significant spatial relationships. High degree of user interactivity is a general requirement to map displays intended to support spatial thinking? i.e. hypothesis generation, data analysis, and decision making. Examples of possible user interactions include:

• Dynamic linking of maps with other types of graphical displays by so-called rushing? objects selected in one of the displays are simultaneously highlighted in all of them;
• Tools for changing in real time parameters of conventional cartographic methods such as the classes in a choropleth map;
• Various interactive devices for controlling map animation

Still, interactive techniques and tools can support information exploration and knowledge construction only when users are able to properly utilize these instruments. User studies demonstrate that effective use of the novel techniques requires learning of the new concepts and ideas. Users are able to understand and adopt the new ideas concerning map interactivity and manipulability. However, these ideas needed to be appropriately introduced; people could not grasp them just from the appearance of the maps and controls. Our Spatial Decision Support Team (SPADE) at the Fraunhofer Institute for Autonomous Intelligent Systems designs and implements novel visualization techniques to support exploratory data analysis and decision making in a spatial context. In particular, we have analyzed traditional methods of graphical and cartographical data representation, revealed their strong and weak sides, and found ways to enhance their strengths and compensate for weaknesses by adding interactivity and dynamics. We combine cartographic visualization with other methods of graphical data representation and data analysis methods from other disciplines, such as statistics and data mining. To support multi-criteria decision making, we combine established techniques for multi-criteria decision support with interactive maps and graphs and invent our own methods, highly interactive and visual. We suggest a range of techniques for decision support accommodating various styles of decision-making. The general topic of the proposed tutorial is visualization of spatial data as a tool for exploratory data analysis, problem solving, and decision-making. The tutorial will be based on the CommonGIS system developed by the SPADE team. The system is available for free use for research and educational purposes. The participants will receive CD-ROMs containing fully functional CommonGIS system and a variety of demonstration projects. Most of the tools considered in the tutorial are unique for CommonGIS since this system have been specially designed for the most efficient support of EDA and decision-making. However, the experience gained in the tutorial can be also utilised in work with commercially available software. In many cases, operations that are effectively “packed?in CommonGIS into a single tool may be done in other packages through sequences of data transformation, calculation, and visualization operations. Some recommendations concerning data exploration and decision making with the use of available commercial software will be given. The tutorial will be held in 1.5 hours and consists of a lecture combined with demonstration. A collection of slides is available at http://www.commongis.com/tutorial/tutorial-ppt.zip

The Star Trek Tricorder for Diagnosis and Treatment is my vision of the medicine of the future. I will be presenting my past and current work towards this long term goal. This work includes volume rendering algorithms and hardware, volume analysis such as segmentation, centerline and skeleton computation as well as computer aided detection of colon polyps and spinal cord fluid atrophy. The interesting tools available to do the image analysis are databases and a large Beowulf cluster.

We propose a simple lighting model to incorporate subsurface scattering effects within the local illumination framework. Subsurface scattering is relatively local due to its exponential falloff and has little effect on the appearance of neighboring objects. These observations have motivated us to approximate the BSSRDF model and to model subsurface scattering effects by using only local illumination. Our model is able to capture the most important features of subsurface scattering: reflection and transmission due to multiple scattering. In our approach we build the neighborhood information as a preprocess and modify the traditional local illumination model into a run-time two-stage process. In the first stage we compute the reflection and transmission of light on the surface. The second stage involves bleeding the scattering effects from a vertex's neighborhood to produce the final result. We then show how to merge the run-time two-stage process into a run-time single-stage process using pre-computed integral. The complexity of our run-time algorithm is $O(N)$, where $N$ is the number of vertices. Using this approach, we achieve interactive frame rates with about one to two orders of magnitude speedup compared with the state-of-the-art methods.

We present a versatile and complete free-form shape modeling framework for point-sampled geometry. By combining unstructured point clouds with the implicit surface definition of the moving least squares approximation, we obtain a hybrid geometry representation that allows us to exploit the advantages of implicit and parametric surface models. Based on this representation we introduce a shape modeling system that enables the designer to perform large constrained deformations as well as boolean operations on arbitrarily shaped objects. Due to minimum consistency requirements, point-sampled surfaces can easily be re-structured on the fly to support extreme geometric deformations during interactive editing. In addition, we show that strict topology control is possible and sharp features can be generated and preserved on point-sampled objects. We demonstrate the effectiveness of our system on a large set of input models, including noisy range scans, irregular point clouds, and sparsely as well as densely sampled models.

We present a comparative study of user performance on tasks involving navigation, visual search, and geometric manipulation, in a map-based battlefield visualization virtual environment (VE). Specifically, our experiment compared user performance of the same tasks across four different VE platforms: desktop, cave, workbench, and wall. Independent variables were platform type, stereopsis (stereo, mono), movement control mode (rate, position), and frame of reference (egocentric, exocentric). Overall results showed that users performed tasks fastest using the desktop and slowest using the workbench. Other results are detailed below. Notable is that we designed our task in an application context, with tasking much closer to how users would actually use a real-world battlefield visualization system. This is very uncommon for comparative studies, which are usually designed with abstract tasks to minimize variance. This is, we believe, one of the first and most complex studies to comparatively examine, in an application context, this many key variables affecting VE user interface design.

I present an overview of several projects at GMU: Graphical Simulation of Fluids, Knee Surgery Assistance System (Funded by Edward MacMahon, M.D.), DEVISE (Funded by US Dept. of Education), MUVEs (Funded by NSF), and Atomic Graphics Pipeline(Proposal to NSF). Some descriptions and images are at: http://www.cs.gmu.edu/~jchen/exhibit.html

1. Real-time Simulation of Fluid Behavior
   
   Graphical simulation of realistic fluid behavior in real time is challenging. We introduce our methods and results in achieving both realistic and real time dust and water behaviors. The methods include using simplified computational fluid dynamics, physics-based modeling, and particle systems. The results are useful in training, education, and entertainment.
2. Knee Surgery Assistance System
   
   A knee surgery assistance system is constructed that includes generating patient-specific 3D knee models from patient?s magnetic resonant images (MRIs), simulating knee motion on the patient specific knee model, and visualizing biomechanical information on the model. The system supports virtual surgery study and practices. Our work is the first effort in integrating 3D reconstruction, motion simulation, and biomechanical visualization into one system.

The Naval Research Laboratory is the Navy's corporate laboratory. NRL conducts a broad-based multidisciplinary program of scientific research and advanced technological development directed toward maritime applications. The Virtual Reality Lab at NRL conducts research and development in emerging virtual and augmented reality technologies to advance Naval warfighting capabilities. Currently, our two main thrusts are the Battlefield Augmented Reality System and multi-modal interaction techniques for virtual and augmented reality. The BARS project examines how three-dimensional strategic and tactical information can be transferred between a command center and individual warfighers who are operating in an urban environment. It is a multi-disciplinary research project that encompasses a number of research and technical issues. These include the development of: (i) novel user interfaces; (ii) new interaction methods; (iii) an interactive, scalable three-dimensional environment; (iv) tracking and registration systems of sufficient accuracy; (v) a prototype demonstration system. Our work on multi-modal interfaces for VR aims to statistically integrate speech recognition, gesture recognition, terrain feature recognition, and 3D virtual reality technology into an extensible, open architecture. Potential applications include any virtual environment with operational significance. Specifically planned applications include battlefield visualization systems and the Battlefield Augmented Reality System (BARS).