|Supervisor:||Dr.-Ing. Jens Bialkowski|
|Faculty:||Prof. Dr.-Ing. André Kaup|
Based on the video coding standard H.263 a development of H.264 as its successor standard had been invoked several years ago. H.264 has been accepted in April 2003 as new international standard as result of the investigations of the Joint Video Team (JVT) consisting of video coding experts of both the ITU and ISO/IEC. As a consequence of the joint work, it is also adopted by ISO/IEC as MPEG-4 Layer 10 Advanced Video Coding. Comparing it with its precessor H.263 or other standards it achieves an excellent coding efficiency, i.e. lower bitrate at constant quality. This makes it attractive for a broad variety of consumer devices and applications in the near future. On the other hand, there exists a great amount of video material in H.263 that consequently must be transcoded to H.264.
In several discussions and investigations of both standards no other way than fully decoding the H.263-coded video and then newly encoding it has been found. Two main aspects exist, that lead to the conclusion, that it does not make sense to fully reencode the input image in all scenarios. First, H.264 get its coding efficiency by a large amount of prediction modes that are difficult to estimate even on modern high end computers. Second, during encoding of the H.263 input bitstream several aspects of the video frame have been selected, which may be used to decrease coding complexity for H.264. A transcoding algorithm should therefore use this side information to decrease coding complexity by limiting the search parameters for reencoding as far as possible.
The thesis work may start by seeking the detailed differences regarding the motion compensation by studiing the standard texts of H.263 and H.264. An investigation of the motin space extracted from the standards software implementation shall then lead to a model for a transcoding scheme. This developed scheme is then to be implemented and analysed at the standards software of H.264. Main target of this implementation is to achieve a high complexity reduction at possibly constant effieciency in terms of coding output.
Basic Knowledge:Video Compression; C-Programming