At low bit rates, most video coders drop frames from the original video. Motion-compensated interpolation (MCI) requires no additional bandwidth and can interpolate the missing frames. Existing MCI methods, developed for uncompressed video, are computationally demanding and require additional motion estimation (ME). By exploiting the motion field already present in the coded video, the MCI algorithms proposed in this paper eliminate or reduce extra ME with negligible performance loss. The algorithms are applied to original and coded video to verify their usefulness.
展开▼
