| Summary: | 碩士 === 國立臺灣海洋大學 === 資訊工程學系 === 97 === As wireless communication network and portable mobile devices getting popular, users can access multimedia without the constraint of locations. When a large number of users access the same video, using a video transcoder can reduce the overload of a server. Video transcoder generates appropriate bitstreams to meet the user’s devices without performing repetition operations of re-encoding.
When video transcoder generates a down-sizing bitstream, it needs to solve the problem of merging macroblock types. This thesis aims at the down-scaling problem and it proposes a fast homogeneous video transcoder. In the case that both the source and target bitstreams use the H.264 standard, server can use the network bandwidth efficiently by taking the advantage that H.264 has the much better performance in terms of image quality and bit-rate. However, H.264 has the disadvantage of high computational complexity for video encoding. In order to decrease the encoding time, a transcoder must eliminate a large amount of motion estimation and motion type selection in the process of encoding. In this thesis, we use the decoded bit stream to estimate the down-scaled motion vectors to speed up video transcoding.
Keywords: H.264/AVC , video transcoding , down-sizing method
|
|---|
