| Summary: | 碩士 === 國立中正大學 === 電機工程所 === 97 === Scalable video coding (SVC) have been developed over twenty years, and the new standard H.264SE supports spatial, temporal and quality scalabilities. But SVC applications are seldom realized in the daily life. The main problem comes from the high complexity which is not acceptable in hardware implementation. So we propose some methods to solve the problems in spatial scalable video coding in this thesis.
To implement the spatial scalability, the inter-layer predictions increase huge complexity for encoding the high resolution layers. Especially, the inter-layer residual prediction and the motion prediction make the complexity of mode decision four times compared to conventional H.264 encoding. The bandwidth is another problem that should be taken into consideration for spatial scalable coding. Usually, the bitstream of the high resolution layer includes bitstream of the associated low resolution layers. Thus, the bitstream loading becomes heavier when more spatial layers are encoded together. In the thesis, we analyze the coding performance of spatial SVC and the complexity of the inter-layer prediction, then some methods are proposed to reduce the complexity of spatial SVC.
The high resolution layer is able to provide accurate information to predict the corresponding low resolution layer. Thus in the proposed methods, the information of the high resolution layer is used to predict the low resolution layer. Besides, we consider the bitstream embedding issue and try to find a balance between the coding performance and complexity. Finally, some simulation results are analyzed and discussed.
|
|---|
