| 摘要 |
In video compression systems, a system and method for reducing visible artifacts in video coding using multiple reference pictures are provided. A first GOP characterized by a“bbIbbBbbBbbB”structure and a second GOP characterized by“bbIbbQbbMbbQ”GOP structure may be used to reduce“I”frame clicking. The“B,”“b,”“Q,”and“M”pictures may be inter-coded by using multiple prediction dependencies, including dependency to a next GOP“I”picture, to smooth out discontinuities across GOPs in a video sequence. Prediction dependencies may be based on a weighing scheme where assigned weights are a function of the temporal distance between reference pictures and the picture being encoded. A first video compression system may be used for the first GOP and may encode“I,”“B,”and“b”pictures. A second video compression system may be used for the second GOP and may encode“I,”“Q,”“M,”and“b”pictures.
|
