Restricting inter-layer prediction based on a maximum number of motion-compensated layers in high efficiency video coding (HEVC) extensions
摘要
說明書
In another embodiment, a layer that doesn't require full reconstruction also doesn't need to reconstruct the residue. Therefore, inter-layer residue prediction from this layer can be disabled and the decoding processes used to derive the residue including de-quantization, inverse transform etc. are not needed.
In another embodiment, the above restrictions of inter-layer prediction may be generalized to be applied within each access unit only. For example, the bitstream, which may contain multiple layers, may be restricted such that in decoding each access unit, motion compensation is used for the reconstruction of at most two layer representations in the access unit. This approach may also be referred to as an access-unit-level two-loop approach. The above approaches, through various restrictions of inter-layer prediction, may all be referred to as two-loop approaches.
In another embodiment, the above approaches may be further generalized to be layer-level N-loop approaches, wherein N could be 2, 3, 4, etc., and the bitstream, which may contain multiple layers, is restricted such that motion compensation is used to decode at most N layers. In another embodiment, the above approaches may be further generalized to be access-unit-level N-loop approaches, wherein N could be 2, 3, 4, etc., wherein the bitstream, which may contain multiple layers, is restricted such that in decoding each access unit, motion compensation is used for at most N layer representations in the access unit. Additionally, the decision about using inter-layer prediction can be dependent on the codecs that are used to code previous layers.
