| 摘要 |
In motion vector competition (MVCOMP) video encoding schemes, the residual motion vector and the index of the chosen motion value predictor vector are encoded for each block. The number of bits required to encode the index may be reduced when some of the predictors are the same or when it is possible to eliminate candidates from knowledge of the motion vector residual and the encoderâ s cost optimization criterion. This bit-reducing method is however very vulnerable to transmission errors because the number of index encoding bits depends on the motion vector predictor values. To counteract this problem entropic encoding is used for encoding the residual motion vector and the index when bitstream losses or motion activity is low, but when losses or motion activity are higher a more robust but less efficient encoding scheme such as unary encoding is used, the selected encoding mode being transmitted as a flag bit in the bitstream. |
