Modeling the dynamics and structure of the Earth&rsquo;s interior now routinely involves massively distributed computational techniques, which makes it feasible to study time-dependent processes in the 3D Earth. 

Accurate, high-resolution models require the use of distributed simulations that run on, at least, moderately large PC clusters and produce large amounts of data on the order of terabytes distributed across the cluster. Visualizing such large data sets efficiently necessitates the use of the same type and magnitude of resources employed by the simulation. Generic, distributed volumetric rendering methods that produce high-quality monoscopic and stereoscopic visualizations currently exist, but rely on a different distributed data layout than is produced during simulation. 

This presents a challenge during the visualization process because an expensive data gather and redistribution stage is required before the distributed volume visualization algorithm can operate. 

We will compare different general purpose techniques and tools for visualizing volumetric data sets that are widely used in the field of scientific visualization, and propose a new approach that eliminates the data gather and redistribution stage by working directly on the data as distributed by, e.g., a seismic wave propagation simulation.

Distributed Volume Rendering of Global Models of Seismic Wave Propagation

A 50 megapixel display wall is under construction at the Cecil H. &amp; Ida M. Green Institute of Geophysics and Planetary Physics (IGPP) for the display of multiple interactive 3D visualizations of various geophysical datasets. This system is designed through collaboration between major NSF funded projects such as OptIPuter and USArray (Earthscope), and will allow researchers to visually analyze data and present results at extremely high resolution. 

The OptIPuter project leverages the capabilities of dedicated optical networks that interconnect instruments, processors, computer storage and visualization resources to aid in Earth Sciences research. This system comprises a cluster of seven Apple Power Mac G5 machines and twelve Apple 30&rdquo; LCD screens (of maximum resolution 2560 x1600 each) tiled to form a 4x3 array and will be the first Apple-driven tiled display to our knowledge. 

The Antelope software will be used for seismic data monitoring and archiving along with web-based analytical tools developed at the Array Network Facility (ANF) at IGPP. 

OptIPuter software (developed by the Electronic Visualization Laboratory) such as JuxtaView (an image viewer for interacting with remotely located extremely high resolution 2D images) and Vol-a-Tile (interactive volume rendering software allowing navigation into gigabyte-sized seismic volumes) will also be used. 

Interactive visualizations created by scientists at IGPP that overlay heterogeneous datasets such as seismic profiles, geology strata, earthquake locations, bathymetry and high resolution satellite imagery and aerial photos, using the Fledermaus software will also be viewed. 

The configuration of each cluster node is: dual CPU 2.5 GHz PowerPC G5, 8 GB RAM, 500 GB disk space, NVIDIA Ultra 6800 GeForce card, and the nodes are interconnected over gigabit Ethernet. 

This system will also be part of the OptIPuter infrastructure, with fiber connections to the OptIPuter CAVEwave on the National Lambda Rail. The design of the system is based on the Geowall-2 class of displays and the OptIPuter Visualization Cluster at Scripps.

High resolution visualization of USArray data on a 50 megapixel display using OptIPuter technologies

A primary need for studies of sediment and rock cores is an integrated environment for visual core description. Digital line-scan images of split-core surfaces should be the fundamental template for all sediment descriptive work, including annotations about structures, lithologic variation, macroscopic grain size variation, bioturbation intensity, chemical composition, and micropaleontology, among other features. Hard-rock visual core descriptions have corresponding requirements. The integration of core-section images with discrete data streams and nested images provide a robust approach to the description of sediment and rock cores. This project provides for the real-time and/or simultaneous display of multiple integrated databases, with all the data rectified (co-registered) to the fundamental template of the core image. Virtual initial core description (ICD) sheets are created on-the-fly as data are acquired. The visualization tool enables rapid multidisciplinary interpretation during the ICD process. Such interpretation is commonly limited to the final step before publication of study results. 

A database exists to contain information on submitted data, including multi-sensor core-logger data, high resolution scanned images of split core sections, text annotation and photomicrographs of smear slides of lithologic constituents and microfossils. A prototype environment for working with the high resolution data is the Personal GeoWall-2, a single computer used to drive 6 tiled LCD screens and a pair of projectors. The combination allows one to juxtapose high resolution core images and sensor data next to fully stereoscopic visualizations.

CoreWall: A Visualization Environment for the Analysis of Lake and Ocean Cores

My first experience with the web started as a Research Assistant for the webgroup at the Electronic Visualization Laboratory (EVL). While studying virtual reality and computer graphics, web technologies were introduced. As time progressed, I realized that my interest was leading me to build art projects on the web. 

The concept of emergent behavior was then introduced which led to ideas of using a social system to fuel visualizations over the internet. The goal was to make a visualization that was organic and behaved as if it had a life of its own, but it was also important to engage the audience. Adopting the notion of the man behind the machine transferred the use of emergent behavior to using human interaction to compose the art.

community

Creating credible virtual reality (VR) computer-generated worlds requires constant updating of the images in all displays to have the correct perspective for the user. To achieve this, the computer must know the exact position and orientation of the user&rsquo;s head. Examples of current techniques for addressing head tracking include magnetic and acousto-inertial trackers, both requiring the user to wear clumsy head-mounted sensors with transmitters and/or wires. 

This thesis describes a two-camera, video-based, tetherless 3D head position tracker system specifically targeted for both autostereoscopic displays and projection-based virtual reality systems. The user does not need to wear any sensors or markers. 

The head position technique is implemented using Artificial Neural Networks (ANN), allowing the detection and recognition of upright, tilted, frontal and non-frontal faces in the midst of visually cluttered environments. In developing a video-based object detector using machine learning, three main sub-problems arise: first, images of objects such as faces vary considerably with lighting, occlusion, pose, facial expression, and identity. Second, the system has to deal with all the variations in distinguishing objects (faces) from non-objects (non-faces). Third, the system has to recognize a target face from other possible faces so it can identify the correct user to track. 

This thesis introduces some solutions to problems in the face detection / recognition domain. For example, it discusses several Neural Networks (NN) per left and right channel, one for recognition, one for detection and one for tracking; real-time NN face and background training (a novel solution which allows new users to spend only two minutes training before being able to use the system); infrared (IR) illumination (to further reduce image dependency cause by room lighting variation) and global image equalization (in place of current trend of local equalization); algorithms highly tuned for the Intel Pentium IV vector processor; and a prediction module to achieve faster frame rates once a face is been recognized. The goal is to reach real-time tracking, in our case 30 frames per second (fps), at 640 by 480 video-image resolution. The system has been evaluated on an ongoing autostereoscopic Varrier&trade; display project achieving 30 frames per second (fps) at 320x240 video-image resolution and 90% tracking position rate. In addition, this dissertation also includes a previous novel work in face detection using LArge Scale Memory STorage And Retrieval (LAMSTAR) neural network from which the current 3D tracker system is derived.

Schematic of Personal Varrier Autostereoscopic Display with Camera Tracking System

Real-time 3D Head Position Tracker System with Stereo Cameras Using a Face Recognition Neural Network

We present the current state of Vol-a-Tile, an interactive tool for exploring large volumetric data on scalable tiled displays. Vol-a-Tile presents a variety of features employed by scientists at the Scripps Institution of Oceanography on data collected from the Anatomy of a Ridge-Axis Discontinuity seismic experiment. Hardware texture mapping and level of-detail techniques provide interactivity. A high-performance network protocol is used to connect remote data sources over high-bandwidth photonic networks.

Vol-a-Tile displaying the ARAD seismic experiment dataset

Vol-a-Tile - a Tool for Interactive Exploration of Large Volumetric Data on Scalable Tiled Displays

We envision situation-rooms and research laboratories in which all the walls are made from seamless ultra-high-resolution displays fed by data streamed over ultra-high-speed networks, from distantly located visualization and storage servers. It will allow local and distributed groups of researchers to work together on large amounts of distributed heterogeneous datasets. 

From our prior work on the Continuum, we have learned that it is crucial for collaborators to have both local control (e.g. on a tablet or laptop) and the casual ability to share their work and see what others are working on (e.g. on a large tile display). We are taking the next steps toward this vision by building LambdaVision - an 11x5 tiled display with a total resolution of 100 megapixels and developing SAGE, the Scalable Adaptive Graphics Environment. 

SAGE will allow the seamless display of various networked applications over the whole display. Each visualization application (such as 3D rendering, remote desktop, video streams, 2D maps) streams its rendered pixels (or primitives) to a virtual high-resolution frame buffer, allowing for any given layout onto the display. 

Computing platforms typically provide some notion of remote desktop access. Tools such as VNC or Microsoft&rsquo;s Remote Desktop were designed to transmit screens of single desktops to remote computers over slow networks. They are designed to operate on event triggered streaming mechanisms that are not suitable for real-time streaming of scientific visualization or collaborative applications. Flexible scalable graphics systems such as Chromium and Aura are designed for distributing visualization to and from cluster driven tiled-displays. Applications have to be specifically written with the Aura API to enable them to transmit their graphics and Chromium was not designed to stream graphics over wide-area networks. 

IBM&rsquo;s Scalable Graphics Engine is also a hardware-based approach that allows for reception of pixels streamed over networks and can drive an array of displays synchronously. Currently, it is limited to a 16-port output and does not support protocols for streaming over wide-area networks. In all these approaches, it is difficult for users to add their own video sources, compression modules or network protocols to the systems, which are neededfor high bandwidth wide-area networks.

Test application running in the SAGE environment

SAGE: the Scalable Adaptive Graphics Environment

In electronically mediated distance collaborations involving scientific data, there is often the need to stream the graphical output of individual computers or entire visualization clusters to remote displays. 

This paper presents TeraVision as a scalable platform-independent solution which is capable of transmitting multiple synchronized high-resolution video streams between single workstations and/or clusters without requiring any modifications to be made to the source or destination machines. 

Issues addressed include: how to synchronize individual video streams to form a single larger stream; how to scale and route streams generated by an array of MxN nodes to fit a XxY display; and how TeraVision exploits a variety of transport protocols. 

Results from experiments conducted over gigabit local-area networks and wide-area networks (between Chicago and Amsterdam), are presented. Finally, we propose the Scalable Adaptive Graphics Environment (SAGE) - an architecture to support future collaborative visualization environments with potentially billions of pixels.

EVL&rsquo;s 6400 x 3072 pixel tiled display with streamed images from laptop

TeraVision: a Distributed, Scalable, High Resolution Graphics Streaming System

JuxtaView is a cluster-based application for viewing ultra-high-resolution images on scalable tiled displays. We present in JuxtaView, a new parallel computing and distributed memory approach for out-of-core montage visualization, using LambdaRAM, a software-based network level cache system. 

The ultimate goal of JuxtaView is to enable a user to interactively roam through potentially terabytes of distributed, spatially referenced image data such as those from electron microscopes, satellites and aerial photographs. 

In working towards this goal, we describe our first prototype implemented over a local area network, where the image is distributed using LambdaRAM, on the memory of all nodes of a PC cluster driving a tiled display wall. Aggressive pre-fetching schemes employed by LambdaRAM help to reduce latency involved in remote memory access. 

We compare LambdaRAM with a more traditional memory mapped file approach for out-of-core visualization.

JuxtaView on GeoWall2

JuxtaView - a Tool for Interactive Visualization of Large Imagery on Scalable Tiled Displays

This paper discusses design issues for enhancing awareness among distributed teams in the display-rich advanced collaboration environments. We  conducted an exploratory design study of nine groups performing a set of collaborative tasks using a variety of advanced collaboration and display technologies. The result showed that group interaction and awareness was improved by the use of multiple public displays due to information visibility to all members. While maximized visibility supported work activity awareness, it also revealed a need to support shared resource awareness and more importantly task awareness for effective group coordination.

Continuum Amplified Collaboration Environment built at the Electronic Visualization Laboratory at UIC, to facilitate an explorative usability study.

Design of Awareness Interface for Distributed Teams in Display-rich Advanced Collaboration Environment

Practicing Seismology in the Fifth-grade Clasroom: Technology in the Service of Developing a Community of Learners

The haptics-based virtual reality periodontal training simulator project focuses upon the research and development of a prototype dental simulator for training of periodontal procedures. By the use of virtual reality and haptics technology, the periodontal simulator allows trainees to learn performing diagnosis and treatment of periodontal diseases by visualizing a three-dimensional virtual human mouth and feeling real tactile sensations while touching the surface of teeth, gingiva, bone and calculi with virtual dental instruments. 

Since periodontics requires dentists to depend primarily on tactile sensations to perform diagnostic and surgical procedures, the use of haptics is unquestionably crucial for a realistic periodontal simulator. 

The haptics-based virtual reality periodontal training simulator has been validated by experiments conducted by the College of Dentistry with faculty members and dental students, which demonstrates the scientific contribution and usefulness of the simulator as a vital part of the curriculum of the Department of Periodontics at UIC. 

The use of medical simulators has proved to increase patient safety and reduce risk associated with human errors in hospitals by allowing medical students to develop skills more efficiently in a shorter period of time. Even though medical simulators are currently being developed by a large number of universities and medical companies, the field of dental simulation has not been well exploited yet. This thesis focuses on the research and development of a haptics-based dental simulator specially designed for training and performance evaluation of dental and hygiene students and practicing professionals in Periodontics. The simulator is the result of a multidisciplinary project developed by the Electronic Visualization Laboratory of the Department of Computer Science, the Industrial Virtual Reality Institute of the Department of Mechanical and Industrial Engineering, and the Department of Periodontics at the University of Illinois at Chicago. The goal of this collaboration between the College of Engineering and the College of Dentistry at UIC is the implementation of a dental simulator to help increase the quality of training for students and future practitioners in the field of periodontics. 

The simulator allows trainees to develop the skills needed to differentiate pathological and normal conditions, as well as to diagnose and treat periodontal diseases. Trainees can learn by feeling tactile sensations as they &ldquo;touch&rdquo; a computer-generated 3 dimensional (3D) model of a human upper and lower dental arches along with various oral components: teeth, gingiva and bone with a haptic device. The simulator also facilitates the learning of proper selection and manipulation of dental instruments to perform
periodontal procedures. 

This thesis is organized in six chapters. Chapter 2 presents an overview of the current commercially-available medical simulators, followed by manikin-based and haptics-based dental simulators previously developed. Chapter 3 gives background information about the diagnosis and treatment of most common periodontal diseases and proposes the use of haptics-based dental simulators to facilitate the learning of these activities. Chapter 4 shows the implementation details and the graphical user interface of the virtual reality periodontal simulator. Chapter 5 describes the use of the simulator for periodontal training, and reports the results obtained after its validation by the College of Dentistry at UIC. Finally, chapter 6 outlines the conclusions of this work and suggests future research and development.

Simulation of Calculi being removed with scaler

Haptics-Based Virtual Reality Periodontal Training Simulator

The current implementation of TCP for the Internet is not efficient when used for Mobile Ad hoc Networks (MANETs). This is because TCP assumes that all packet losses are caused by congestion, whereas transmission errors are a main reason for packet losses in wireless networks. 

To ameliorate this situation and increase performance, we propose a method of using multi-metric parameters to distinguish the causes for packet losses and use Colored Petri Net to analyze the revised protocol. We call this TCP-MEDX (Mobile Error Detection eXtension). 

TCP-MEDX has two characteristics: Firstly, it is energy-efficient because this solution is only initiated when a packet loss is detected. This characteristic is very important for MANETs because of its limited power source; secondly, our approach removes negative effects caused by asymmetry in wireless links, thus improving correctness in determining causes for packet losses. 

Our simulation results using Design / CPN show that the proposed approach increases throughput and reduces propagation delay compared with standard TCP.

Energy-Efficient Method to Improve TCP Performance for MANETs

ACM Copyright rules prohibit putting the PDF on our website. Please follow the links below to review or purchase the article. 

<a href="http://doi.acm.org/10.1145/1012037.1012059">&ldquo;Designing Cranial Implants in a Haptic Augmented Reality Environment&rdquo; - CAMC August 2004.</a> 

<a href="http://www.acm.org/publications/policies/copyright_policy">Copyright 2004 by ACM, Inc. - Copyright Restrictions on use can be viewed here.</a> 

Repairing severe human skull injuries requires customized cranial implants. Traditionally, medical sculptors have employed their anatomical modeling expertise to sculpt prosthetic implants using clay and wax. However, even with the aid of automated manufacturing techniques, the design process remains expensive in terms of labor, materials, and money. Any modification to a sculpted implant requires fabricating a new model from scratch. Techniques developed at the University of Illinois at Chicago (UIC) in 1996 have greatly improved this practice. Virtual reality research now aims to augment these tools and methods within a networked digital medium. 

Closing large cranial defects offers patients therapeutic benefits, including restoring the shape of the head, protecting vital brain tissue, minimizing pain, reducing operating and recovery time, and in some cases even improving cognitive abilities. Unfortunately, many factors limit cranial implant availability. First among them, insurance companies often do not support this form of reconstruction due to the high cost of labor and manufacturing. Second, because only neurosurgeons and medical modelers have the specialized anatomical and technical knowledge to do the work, assembling the necessary expertise is difficult; travel expenses for both patients and specialists further increase the overall cost. Finally, the process of solidifying implant material exudes extreme temperatures. Exposing the brain to these temperatures damages tissue, so presurgical implant design is often critical to reducing patient risk. Traditional cranial implant fabrication and surgical placement methods depend on subjective skills and procedures.

A medical sculptor &ldquo;touches&rdquo; a cranial defect while sitting at PARIS

Designing Cranial Implants in a Haptic Augmented Reality Environment

Sort-Last is a computer graphics technique for rendering extremely large data sets on clusters of computers. Sort-Last works by dividing the data-set into even-sized chunks for parallel rendering and then composing the images to form the final result. 

Since sort-last rendering requires the movement of large amounts of image data between cluster nodes, the network interconnecting the nodes becomes a major bottleneck. 

In this paper, we describe a sort-last rendering system implemented on a cluster of computers whose nodes are connected by an all-optical switch. The rendering system introduces the notion of the Photonic Computing Engine, a computing system built dynamically by using the optical switch to create dedicated network connections between cluster nodes. 

The sort-last volume rendering algorithm was implemented on the Photonic Computing Engine, and its performance is evaluated. Preliminary experiments show that performance is affected by the image composition time and average payload size. In an attempt to stabilize the performance of the system, we have designed a flow control mechanism that uses feedback messages to dynamically adjust the data flow rate within the computing engine.

A Sort-Last Rendering System over an Optical Backplane

Artists, scientists and mathematicians have been collaborating on a variety of projects at the Electronic Visualization Laboratory for over thirty years. In 1989, a ray tracer was created to visualize higher dimensional Julia Sets, involving and contributing to advances in all three fields. 

Background 

The creation of the ray tracer to visualize higher dimensional Julia Sets described here took place in a research environment called the Electronic Visualization Laboratory (EVL). EVL is a shared facility of the School of Art and Design and the Department of Computer Science at the University of Illinois at Chicago. In operation for over 30 years, EVL has two directors, 12 associated faculty and staff, and 50 graduate students. About one-third of the students are pursuing a Master of Fine Arts (MFA) in the School of Art and Design and two-thirds are pursuing either a Master of Science or a Ph.D. in computer science. To our knowledge, this is the longest living program that offers an MFA which is a formal collaboration of art and computer science. EVL delivers art intelligence to science, and science and technology to artists. It systematically teaches the artists the technology and less systematically teaches the art to the computer science students. The artists&rsquo; experience is central to the success of EVL. EVL creates and expands new media and supplies art and science content for those media. 

The economic support of the EVL is based on working with scientists and mathematicians to deliver new visualization techniques and technologies to support scientific and mathematical investigation. The research outlined in this paper would not have been possible without EVL and the expertise of both the computer scientists and artists. 

There has been considerable interest in the visualization of fractals. Their infinite detail combined with repeating forms have intrigued mathematicians and artists and broadly appeal to the general public. The most popular images have been of two-dimensional Mandelbrot and Julia Sets.

A Ray Tracer to Visualize Higher Dimensional Julia Sets

Processing data for print and other static media may employ numerous design, digital and analog steps along the way; but the final product is dated at its release. In contrast, real-time visualization of data employs an active system capable of processing and rendering data as a dynamic presentation. 

Rather than displaying a single snapshot of data, real-time information visualization presents the data the moment it is queried. As it is driven by computers equipped with memory caches, there is the added ability to observe both the immediate effects of input affecting data, or to observe data behavior over a period of time. 

Designing an information visualization backed by digital technology poses a unique set of challenges; but like all design processes, the goal is to make complex data simple and understandable. 

Using computational data to generate computer graphic art work dates back at least three decades and has evolved into a discipline in itself - electronic visualization. The use of massive scientific datasets to generate real-time art is a newly emerging field based largely on the increasing ability to access high-bandwidth networks. 

Since 1997, the Electronic Visualization Laboratory&rsquo;s networking initiatives, including STAR TAP and StarLight1, have facilitated national and international R&amp;D network interconnectivity in support of high-end application, networking, visualization and data mining user communities. Since 1998, EVL has also organized biannual international grid, or iGrid, events where these communities converge to showcase multi-gigabit-enabled applications running over experimental testbeds. 

My interest in developing a real-time network information visualization came in early 2000, when STAR TAP networking engineers requested a graphics-based traffic map for iGrid 2000 to display the in / out bandwidth of multiple networks connected to a central router at the Ameritech Network Access Point (NAP) in downtown Chicago.

&ldquo;Sphere&rdquo; Visualization of Network Traffic from Transitive

Transitive: Immersive Visualization of Real Time Network Transfers

In this report, we introduce a flexible and scalable volume visualization system for visual browsing of spatially and temporally large datasets in distributed computing environments. Our framework leverages advances in high-speed networks for access to shared scalable computation, data storage and graphics. In particular, we assume commodity off the shelf graphics hardware for both, rendering and data processing. We describe the implementation and integration of the various pipelines within our visualization framework and present examples of applications in seismology and biological tomography. The quintessential goal of this research is to create an adaptive visualization framework that is cognizant of the constraints in the various pipelines and accordingly, optimizes the visualization process dynamically.

ARAD image

Volume Rendering of Large Data for Scalable Displays Using Photonic Switching

Repairing severe human skull injuries requires customized cranial implants, and current visualization research aims to develop a new approach to create these implants. 

Following pre-surgical design techniques pioneered at the University of Illinois at Chicago (UIC) in 1996, researchers have developed an immersive cranial implant application incorporating haptic force feedback and augmented reality. The application runs on the Personal Augmented Reality Immersive System (PARIS&trade;), allowing the modeler to see clearly both his hands and the virtual workspace. The strengths of multiple software libraries are maximized to simplify development. This research lays the foundation to eventually replace the traditional modeling and evaluation processes.

The defect tool traces a path for extracting the defect geometry

Pre-surgical Cranial Implant Design using the PARIS&trade; Prototype

The StorySpace project studies the role new interface technologies might play in high school education. Unfortunately, technology often seizes center stage in the classroom, becoming itself the topic of instruction. We argue that learning in general and learning about technology in particular will be most successful when technology is used in the service of learning, rather than learning in the service of technology. Certainly, outside of the classroom, technology is rarely an end in itself;  application gives technology its power. The classroom should be no different.

StorySpace

StorySpace: Technology Supporting Reflection, Expression and Discourse in Classroom Narrative

This paper describes the development of a generic framework for implementing realistic cross-platform haptic virtual reality applications. Currently, freely available Software Development Kits (SDKs) deal with a single Haptic Interaction Point (HIP), i.e. the tip of the haptic probe. However, many applications as path planning, virtual assembly, medical or dental simulations, as well as scientific exploration require object-object interactions, meaning any part of the complex 3D object attached to the probe collides with the other objects in the virtual scene. Collision detections and penetration depths between 3D objects must be quickly computed to generate forces to be displayed by the haptic device. In these circumstances, implementation of haptic applications is very challenging when the numbers, stiffness and/or complexity of objects in the scene are considerable, mainly because of high update rates needed to avoid instabilities of the system. The proposed framework meets this high requirement and provides a high-performance test bed for further research in algorithms for collision detection and generation of haptic forces.