抄録
We previously proposed a lossless coding method for color video signals in RGB 4:4:4 format.
In this method, a color signal of the target pixel is predicted using not only the same color of the current
and motion-compensated past reference frames, but also the other colors of the current frame if they are
available. In this paper, use of another reference frame, whose color signal is extrapolated from the previous
frame based on pixel-wise motion estimation, is examined to improve prediction accuracy.