1993 Graduate Student Research Conference Program (GSRC93)


Session 1:Computer Graphics and Image Analysis


Hierarchical Shape Description and Recognition

Jiangying Zhou and Theo Pavlidis

A critical problem in character recognition is the construction of a description of character shapes. The major challenge we face in solving this problem is to find a description that is complete and intrinsic for the recognition of the shape. We think that a description which is constructed a priori to the knowledge of a character is probably perceptually insufficient and inevitably entails some degree of uncertainty in the recognition of the character. In this talk we will present a multi-level representation scheme, in which it is possible to defer an interpretation about a particular part of a shape if there is not sufficient evidence, and the description is completed in the subsequent levels.

Volume Sampled Voxelization of Geometric Primitives

Sidney W. Wang and Arie Kaufman

We present a 3D antialiasing algorithm for voxel-based geometric models. The technique band-limits the continuous object before sampling it at the desired 3D raster resolution. The algorithm not only creates voxel models which are free from object space aliasing, but it also incorporates the image space antialiasing information as part of the view independent voxel model. The resulting alias-free voxel models have been used to model synthetic scenes, for discrete ray tracing applications. The discrete ray-traced image is superior in quality to the image generated with a conventional surface-based ray tracer, since silhouettes of objects, shadows, and reflections appear smooth (jaggy-less).

Global Illumination Models for Volume Rendering

Lisa Sobierajski and Arie Kaufman

Global illumination models have been used extensively in the rendering of classical geometric objects, but have not been rigorously investigated in the area of volume rendering. Volume rendering techniques typically employ only a local illumination model for shading, and therefore produce images without global effects. In this talk the application of global illumination models to volume rendering will be explored. The use of standard global illumination models for rendering isosurfaces contained within scalar fields will be described. A volumetric global illumination model, along with the associated volume rendering algorithm, will also be presented.

Parallel Volume Rendering

Claudio Sliva and Arie Kaufman

Volume rendering is a powerful, but computational intensive, computer graphics technique. In recent years its use has grown from small research applications to industrial medical and geological applications to name a few. Now even PC games use volumetric data. In this talk a new approach to parallel volume rendering, the "pipelined parallel ray caster" will be presented. The PPRC currently runs on networked workstations and the Intel iPSC/860. The algorithm and the implementation will be described along with latest results.

The VolVis Volume Visualization System

Rick Avila and Arie Kaufman

The VolVis volume visualization system has been developed which contains a powerful set of tools for displaying and analyzing three dimensional data. The VolVis system is supported by a generalized abstract model which provides for both geometric and volumetric constructs. VolVis also contains several rendering algorithms including ray tracing of isosurfaces and transparent volumes. In addition, a fast volume rendering algorithm has been developed, which is capable of exploiting existing graphics hardware without placing any viewing restrictions or compromising accuracy. VolVis also includes a volumetric navigation facility, an animation generator, quantitative analysis tools, and a generalized protocol for communicating with 3D input devices.

Session 2: Concurrency and Systems


"Concurrency Factory": A Tool for Design of Distributed Systems

Oleg Sokolsky and Scott Smolka

Communication is the critical part of distributed systems, where several independent devices or processes perform concurrently. "Concurrency Factory" is a software tool that allows to design the communication structure of a distributed system as a hierarchical network of communicating finite-state automata. It also provides facilities for simulation of the resulting system and its verification against a logical or behavioural specification.

The Semantics of State Refinement in Process Algebra

Andrew Uselton and Scott Smolka

We introduce a state refinement operator into BPA with recursive specifications and present a comprehensive technical development of the resulting theory, BPA+SR. Our main technical results are that bisimulation is a congruence in BPA+SR and that guarded recursive specifications have unique solutions. We also have that bisimulation remains a congruence if the merge operator of ACP is added to BPA+SR. This is significant since action refinement, another approach to refinement in process algebra, does not in general preserve semantic equivalences based on interleavings of atomic actions.

An Integrated Approach to Communication Axioms Among Concurrent Processes

Rui Hu and Philip Lewis

Currently different kinds of axiom schemas are used to describe synchronous communication, two-stage shift register communication and virtual circuit communication. Here we define a common model to describe those different communications. The underlying idea of the new model is that the nth receive command gets the value sent by the nth send command. We have the same axiom schema for all communications that share this property. The difference between the communication method is modeled by different termination conditions for the axioms.

Client-Server Scalability and Reliability

Michael Vernick and Gary Schloss

The increase in power and capacity of workstations and personal computers has created a trend in the computer industry where enterprises are moving away from the traditional dumb terminal/main-frame environment and into the client-server environment. In the client-server architecture (CSA), the server provides the computing power for shared data manipulation and access. Our goal is to increase the scalability and reliability of the CSA system. We present a description of a hybrid architecture called HCSA (Hybrid Client-Server Architecture) which is a combination of the CSA and Shared-Disk architectures. We also present the EWS (Exclusive Write Protocol) where clients read directly from disk, bypassing the server, but write through the server.

Electronic Design & Test Automation at Stony Brook

Ramesh Sathianathan and David Smith

The design and test of a VLSI circuit is a complicated process that goes through several phases. At each phase, the design is represented at a different level of abstraction. A suite of CAD tools have been developed at Stony Brook to support the design and test of a circuit, described in the SBL hardware description language, and facilitate the generation and test of its equivalent circuit layout.

Session 3: Multimedia Systems


Virtual Input Devices for 3D Systems

Taosong He and Arie Kaufman

The device unified interface is a generalized and easily expandable protocol for the communication between applications and input devices. The key idea is to unify various device data into the parameters of a so-called ``virtual input device.'' The device information-base, which includes device dependent information, is also incorporated into the virtual input device. Using the device unified interface, system builders are able to design their applications independent of the input devices as well as utilize the capabilities of several devices in the same application.

Geographical Information Systems and Data Structures

Cristian Mata and Joe Mitchell

Geographical information systems (GIS) will become more important as domain independent computing becomes more widespread. Already, applications pertaining to global orientation, positioning, trajectory tracking are in use. One of the problems of GIS is that as the area covered by such a system increases, the complexity of the underlying data structure tends to grow proportinally. E.g. a city map can be kept in the main memory of a present day computer but the entire map of a state cannot. One approach of solving this problem is to reduce the complexity of the map keeping some desired properties. Mathematically, since most maps are modelled as graphs or weighted subdivisions, this means the use of graph spanners. We will present some thoretical results in the field and some practical applications in the computation of shortest paths.

Near-line Storage Systems

Chitra Venkatramani and Tzi-cker Chiueh

With the coming of multi-media applications and applications that incorporate large amounts of data into their computations, the need for a new level in the traditional storage hierarchy has arisen. This is the near-line storage level that provides fast access to massive amounts of data. Mass storage devices constitute storage systems of this kind. A taxonomy of mass storage technologies according to recording medium, recording mechanism and cost/performance characteristics is given. The two most important types of mass storage devices, classified based on the recording medium --- magnetic and optical --- are discussed in detail. Also, an indepth tutorial on related architectural issues for computer scientists is provided.

File System Support for Multimedia Applications

Thirumale Niranjan and Gary Schloss

Multimedia (MM) is an application domain that imposes unique demands on the file underlying system. Previous work assumes that MM is just another special case, which, like many other applications (most notably, database systems), should handle its data I/O by implementing its own file system, usually on a separate disk partition. In this talk we briefly survey the major challenges that generic file systems like UNIX FFS must overcome in order to adequately support MM file storage. Then, we propose and describe a modified file system, which is very predictably called: Multimedia File System (MMFS), that in our view can significantly improve FFS performance in a generic environment of mixed MM and non-MM applications.

Data Models for Multimedia

Michael Wynblatt and Gary Schloss

As multimedia presentations becomes more commonplace, tools are needed to aid in their development. At Stony Brook, we are developing a data model which can describe multimedia's complex characteristics including data stream synchronization and sequencing, intricate display layout, and interactivity. Using a layered design, the model stresses the conceptuual differences between data, data presentation and data manipulation. The model is designed as a conceptual tool, but is intended for extension to a practical development system.

Session 4: Programming Languages


XSB -- Stony Brook Prolog

Terrance Swift and David Warren

The XSB Programming System is a shareware logic programming system that includes full functionality of Prolog. A goal of XSB is to tie together the logical power of Prolog with the declarativeness and persistence of deductive database systems. To this end, XSB extends Prolog by adding tabling and HiLog, two features used by the newest generation of deductive database systems. However XSB's engine executes at the speed of compiled Prolog and evaluates most programs faster than available deductive databases can. Version 1.2 is available by anonymous ftp from sbcs.sunysb.edu.

Determinacy Analysis for Efficient Prolog Execution

Steven Dawson and I. V. Ramakrishnan

In order to realize more fully the potential of Prolog as a declarative programming language, many performance issues need to be addressed. Prolog's ability to simulate non-deterministic computation, while a powerful problem-solving tool, incurs substantial overhead for programs that express determinate computations. A sophisticated program analysis method has been developed at Stony Brook that can reveal at compile time many kinds of determinate computation that have eluded previous approaches. This analysis can benefit both sequential and parallel Prolog implementations.

Trends in Debugging of High-level Programming Languages

Karen Bernstein and Eugene Stark

High-level programming languages abstract the programming process allowing programmers to concentrate on what their program does rather than how it does it. Suprisingly, even though debugging is closely related to programming, debugging for languages such as Prolog and ML is done at a much lower level of abstraction. Debuggers for high-level languages require the programmer to reason explicitly about how the program is executing rather than concentrating on the result. In this talk, I will explain some of the reasons why debugging is difficult and survey some popular proposed approaches to simplifying the debugging process.

Object Oriented Programming

Sergio Martin and Jack Heller

The Object-Oriented paradigm has found a widespread interest in recent years, both at academia and industry. In the context of large software projects requiring industrial-quality software, the Object-Oriented approach is concerned with better ways for software producers to deliver functionality to software consumers. This work proposes an Object Design Representation targeted to express not only conceptual aspects of the object-oriented paradigm, but also some of the implementational characteristics found in object-oriented languages, with particular emphasis on C++.
Return to the GRC home page
If you have problems with this page, please send E-mail to:
evans@cs.sunysb.edu