Parallelism Aware Fast Mode Decision Algorithm for Pre-loaded Spatial SVC Encoding

Scalable Video Coding (SVC) is a new encoding structure presented recently. Comparing to the traditional encoding method, SVC can encode many video with the same content but different quality/size into one bit stream. Users can decode the bit stream into whatever quality they want according to their hardware support and the internet environment. This structure can efficiently gain the suitability of video encoding and decoding.

Due to the high popularity of high resolution videos in the future, encoding such videos may consume lots of time. To speed up SVC encoding, under the scenario that data is pre-loaded, this thesis will try to find out a fast mode decision algorithm which can not only reduce the complexity of encoding but not to harm the quality. The contribution of this thesis can efficiently reduce 75\% of encoding time and achieve speed-up more than 4 times.