Combined adaptive-fixed interpolation with multi-directional filters

Abstract

By optimizing the variable filter coefficients through calculating Wiener–Hopf equations for each picture, the adaptive-interpolated prediction can significantly improve the coding efficiency. However, compared with the fixed coefficient interpolation filtering, this technique adds a huge amount of complexities in implementation. To reduce the complexity of the interpolating unit, this paper proposes a sub-pixel interpolation scheme, which is called combined adaptive-fixed Interpolation (CAFI) with multi-directional filters. Compared evaluations of the hardware complexity as well as the coding efficiency are given. The results show that the CAFI has lower complexity and a stable high coding efficiency for test sequences.