摘要 |
A prediction unit (502) is identified within a picture of a video sequence. The prediction unit (502) is bordered by neighboring prediction units within the picture. A match for the prediction unit is located within a first reference picture. At least a subset of the neighboring prediction units is scanned (604) to identify a motion vector predictor ("MVP") candidate. The MVP candidate of a first type is identified if available, wherein a match for the MVP candidate of the first type is located within the first reference picture. In response to the MVP candidate of the first type being unavailable, the MVP candidate of a second type is identified, wherein a match for the MVP candidate of the second type is located within a second reference picture. An encoder calculates a difference between a motion vector of the MVP candidate and a motion vector of the prediction unit, and encodes the difference and an index to identify the MVP candidate. A decoder decodes the difference, and calculates the motion vector of the prediction unit by adding the difference to the motion vector of the MVP candidate. |