Abstract
Very low bit-rate video coding, using regular shaped patterns tofocus on moving regions in macroblocks, has gained significantattention recently. This paper presents a new real-time paltemselection (RTPS) algorithm using a large codebook of thirty twopattems. The algorithm uses a relevance measurement for all thepattems and a moving region, to eliminate a large number ofirrelevant pattems prior to the actual best likelihood patternselection procedure. Both theoretically and empirically it isproven that not only is the computational complexity of the newalgorithm comparable to the contemporary algorithm that use apattern codebook size of only eight patterns but also the newalgorithm reduces the bit-rate significantly, while maintainingcomparable subjective quality.
| Original language | English |
|---|---|
| Title of host publication | IEEE International Conference on Acoustics, Speech, and Signal Processing |
| Place of Publication | USA |
| Publisher | IEEE |
| Pages | 397-400 |
| Number of pages | 4 |
| Publication status | Published - 2003 |
| Event | IEEE International Conference on Acoustics, Speech, and Signal Processing - Hong Kong, Hong Kong Duration: 06 Apr 2003 → 10 Apr 2003 |
Conference
| Conference | IEEE International Conference on Acoustics, Speech, and Signal Processing |
|---|---|
| Country/Territory | Hong Kong |
| Period | 06/04/03 → 10/04/03 |