Summary: | 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 88 === In this thesis, we present the video coder which achieves various continuous scalabilities simultaneously. By applying the wavelet transform that maps integer to integer in spatial and temporal dimensions, it is possible to perfectly reconstruct the original video sequences. Due to the energy compaction in the temporal domain, using wavelet transform alone does not work well, we replace it by applying the motion estimation at the Low-Low(LL) band to eliminate ringing effects. Following the 3D SPECK algorithm to represent the wavelet (or residual) coefficients based on the bit plane quantization (according to magnitude comparisons), a precise rate control and self-adjusting rate allocations are achieved.
In addition to being rate scalable, the video coder allows
multiresolutional scalabilities in both time and space from
several layered bit-streams. This added functionality along with
many desirable attributes, such as full embeddedness for
progressive transmission, precise rate control for constant
bit-rate traffic, and low complexity for possible software-only
video applications, makes the proposed video coder an attractive
candidate for multimedia applications.
|