A video coding method that derives adjusted motion vectors through offset-based refinement to generate enhanced inter-coded predictions.

Patentradars sammanfattning

The invention relates to inter-frame video coding techniques that process motion vector differences (MVD) to generate updated motion vectors (MV), which are then refined through motion vector offset (MvOffset) calculations. The method produces two sets of prediction blocks—initial predictions from updated MVs and refined predictions from adjusted MVs derived via offset application—enabling more accurate motion-compensated prediction in video compression systems compliant with H.264/AVC or HEVC standards.

AI-genererad och redaktionellt processad.