Dirac Home
Navigation item arrowHome
Navigation item arrowDocumentation
Navigation item arrowDirac Algorithm
Navigation item arrowContents
Navigation item arrowIntroduction
Navigation item arrowArchitecture
Navigation item arrowRDO
Navigation item arrowTransform coding
Navigation item arrowMotion estimation
Navigation item arrowMacroblocks
Navigation item arrowMotion vector coding

SourceForge.net Logo
Valid XHTML 1.1!
The Dirac Algorithm

This document describes a prototype video compression system based on motion-compensated wavelet coding. The system incorporates tools for motion estimation and motion compensation based on overlapped blocks to eliminate block artefacts. Entropy coding is based on adaptive arithmetic coding, and the reference coder is capable of trading off the bit rate of different elements of the bitstream using Rate-Distortion Optimisation techniques. The codec has not been very extensively tested or optimised, but the performance to date suggests that the codec is at least twice as efficient as MPEG-2.

This is a general description of the Dirac algorithm, and some details may have changed since this was written. Refer to the Dirac Specification for the most up-to-date information.

Next: Table of Contents

This document refers to Version 0.9.1 of Dirac

Written by Thomas Davies, BBC R&D. Last revised 6 February 2008