Compensating for drift in the down conversion of high definition sequences to lower resolution sequences

摘要

In a reduced memory MPEG decoder, drift caused by the application of pixel resolution motion vectors during prediction filtering is reduced. Received P and B frame and field DCT coded blocks are down converted. Reference pixels are selected based upon a received motion vector. The selected reference pixels are prediction filtered in order to produce a prediction output. If the motion vector is a pixel resolution motion vector, a drift compensation value is added to the prediction output. Then, the prediction output is added to the down converted P and B frame and field DCT coded blocks in order to form lower resolution reconstructed pixel blocks