Macroblock adaptive frame/field coding architecture for scalable coding

摘要

An open loop encoding architecture encodes a sequence of interlaced video frames at macroblock level. In one aspect, each frame is divided into pairs of macroblocks and the macroblock pairs are encoded as either separate macroblocks or as two fields, depending upon a motion threshold. Predictors for the macroblock pairs may be selected from different frames in the sequence, or from frames of different resolution. In another aspect, a frame may be open loop encoded at field level instead of at macroblock level. A corresponding inverse open loop encoding architecture is used to decode the encoded frames.