| 摘要 |
In order to achieve a constant video quality, the anchor and non-anchor frames of different frame types (I,P,B) are encoded using a different number of bits. However, since video sequences generally contain widely varying picture content and previously coded frames are used to predict a given frame, a suitable assignment of frame target bit rates is hard to determine, especially for non-anchor frames. According to the invention, non-anchor frames are coded using a fixed quantisation parameter. Since the quantisation parameter used for the encoding of non-anchor frames is directly derived from the average quantisation parameter of the previously encoded anchor frame, such approach ensures a constant video quality. Beside of that, the complexity of the rate control startegy is reduced, because no macroblock-level rate control is applied for the encoding of non-anchor frames. <IMAGE> |
