| Permutation Based Video Compression 
Description Video compression has been widely studied and a number of high quality video encoders exist today. The current generation video compression algorithms such as H.264 and Windows Media Video (WMV) are considered to have maximized the compression efficiency. This invention uses permutations to improve the compression efficiency. The proposed encoder generates a sorting permutation of the uncompressed video. When such a sorting permutation is applied, the uncompressed video becomes highly redundant and results in high compression. The cost of communicating such a permutation is, however, very expensive and consumes significant bandwidth/storage. To overcome this limitation, a sorting permutation of the previously coded frame is used. Such a permutation differs from the actual sorting permutation of the current frame, but is close enough to result in highly redundant frames after the transformation. The transformed frame is then encoded using a simple hybrid coding algorithm such as MPEG-1. The proposed method can also be used together with H.264 or WMV. Preliminary results have shown that this permutation based scheme can perform more than four times better than H.264, because it requires only a quarter of the storage capacity or communication bandwidth. The proposed algorithm has substantially lower complexity compared with the current generation algorithms. There is currently no known work on permutation transformations for video compression. And, the use of the permutation of a previously coded video frame instead of the actual sorting permutation is another novel innovation. Potential Applications - Significantly reduce the bandwidth required for applications such as digital TV and DVDs
- Secure video conferencing
- Video compression and communication applications
- Tele medicine
State of Technology Development and Commercialization
Algorithm development is mostly complete. Prototype development is currently underway (60% complete).
(c) Florida Atlantic University All rights reserved.
Updated August 9, 2008 | 
