| 摘要 |
Motion Estimation is one of the most computationally complex processes within a video encoding system. Several Fast Integer Sample Motion Estimation schemes are introduced to reduce complexity of considering integer positions during motion estimation. However, after making such considerations fractional sample motion vector refinement becomes instead the greatest bottleneck for video encoding. Accordingly, algorithms that can reduce such complexity, with little impact in quality, are required and are also disclosed herein.
|
