Past Projects
Past Projects
PowerPoint Presentations:
Research Projects at NSF Multimedia Laboratory (1995-2001)
Video and Image Internet Broadcasting Systems
Real-Time Video Compression
W e have been researching new real-time compression and coding algorithms.
We developed the XYZ algorithm for video compression (Ph.D. Dissertation
of Raymond Westwater, 1996), which is based on the Three-Dimensional Discrete
Cosine Transform. The elegance of the algorithm is its simplicity, which
and high compression while maintaining high quality of video. This algorithm
is well suited for applications such as video-on-demand, videoconferencing,
interactive multimedia, and videophone. We also researched coding systems
for transmission of video on low bit rate communication lines (Ph.D. Dissertation
of Joseph Celi, 1998).
Interactive Progressive Coding Systems
In this area, we are researching new interactive algorithms that allow users
to receive high-quality images and videos over the networks with limited
bandwidth. We have developed a fully workable interactive progressive JPEG
system for coding complex images in applications such as space exploration,
storage and transmission of medical images, Web applications, and others.
The compressed images, stored on the server, are transmitted to users in
several scans. The user can interactively select the parts of the image
that will be sent from the server in the next scan.
Multimedia Broadcasting Over the Internet
We are investigating new techniques for broadcasting audio and video over
the Internet. We have developed a new protocol that provides inexpensive,
efficient, and reliable audio and video broadcasting. Applications include
radio and television broadcasting over the Internet, distance learning,
videoconferencing, real-time distribution of critical data, and others.
Internet Architectures for Distributed Applications
In this area, we have been investigating novel multi-tier Internet architectures
for distributed applications such as Application Service Providers. We have
been investigating various components of these architectures including their
security, reliability, and performance.
Fast Motion Estimation Algorithms
In this area, we have developed and implemented a new fast motion estimation
algorithm for video compression (Master thesis of Joshua Greenberg, 1997).
The algorithm replaces the current motion estimation algorithms applied
in MPEG and H324/H263 standard coding schemes. The complexity of the algorithm
is comparable to other motion estimation techniques, however it shows superior
results in terms of compression ratios and video quality).
Video and Image Retrieval and Indexing Techniques
We have been researching new techniques for video and image retrieval and
indexing that cab be used in designing contemporary database systems. A
new technique, based on histogram functions of the DCT coefficients, is
capable of retrieving images and video stored in compressed form. We have
also been researching new techniques for content-based image and video retrieval
using relevance feedback (Ph.D. Dissertation of Oge Marques, 2001).
Processor Architectures to Support Multimedia
We have been researching new architectures, based on SIMD and parallel processors,
that can effectively support contemporary multimedia applications. We are
exploring both general-purpose processor architectures as well as specialized
architectures for video and image processing.