抄録
The coding efficiency of block-based video coding methods such as MPEG-2 is low in parts where different regions coexist. Segmentation-based methods, therefore, have great potential as next-generation video coding schemes. The segmentation method described here minimizes the generated entropy and is used for an optimal segmentation-based video coding scheme that is adaptive to coding rates. The method achieves optimal segmentation under a rate constraint by es-timating generated bits according to quantization step size and by the use of layered dynamic programming. This leads to higher efficiency compared with the conventional block-based coding for many kinds of video sequence.
