A Foveated Imaging System to Reduce Transmission Bandwidth of Video Images from Remote Camera Systems
Abstract
This is the final progress report for a joint project between Geisler's laboratory at the University of Texas at Austin and OWL Displays. Inc., to develop a real time variable resolution (foveated) imaging system for video communications tasks such as remote control of unmanned vehicles. Although the OWL side of the project might best be described as a modest success, we believe the software and software/hardware integration performed at UT has been a big success. A general purpose real-time linkable library (for Pentium class computers running the Windows95/98/NT OS) has been developed for coding and decoding variable resolution static images and video, both in 8-bit gray scale and 24-bit color. Real-time demonstration executables using the library are currently available at our web site for this project: http://fi.cvis.psv.utexas.edu. We have tested our real time software in conjunction with MPEG (H.263) and shown that it generally produces very substantial bandwidth savings both for I frames and P frames. We have also developed our own real time image compression library which includes fast motion compensation, fast pyramid coding, fast zero-tree coding and arithmetic coding. The foveated imaging software has been successfully interfaced and tested with a 512x512 8-bit b/w camera and two separate eye trackers: an ASL desktop heads-free eye tracker and the OWL/ASL V8 helmet mounted eye tracker.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 12, 1998
- Accession Number
- ADA359067
Entities
People
- H. L. Webb
- Wilson S. Geisler