| 摘要 |
<p>A method and system for re-encoding frames of a digital video stream without changing the previously calculated bit length of the stream. A set of macroblocks at a time are re-encoded (292), the number of resulting bits is calculated (306, 368) and the deviation from an estimated number of bits is calculated (372), either set-by-set or cumulatively. Based on the accuracy (372), of the estimate (422) and the remaining number of sets of macroblocks (424) to re-encode, a correction factor (710) for the quantization level of the next set of macroblocks is chosen from a lookup table. The correction factor is added to the quantization level of the next set of macroblocks (712) and the results stored as the new quantization level of the set (706). This sequential process is done for all sets of macroblocks in the digital stream which are to be recorded according to changed image qualities (706, 708, 710, 712, 714).</p> |
