| Plenty of coding tools are being developed to boost the efficiency of H.264/AVC. Some of these tools are implemented in the key technical area (KTA). The adaptive interpolation filter (AIF), the high precision interpolation filter, and the adaptive loop filter (ALF) have already been introduced in order to provide a more precise reference picture. Other efficient coding methods, such as the motion vector competition and the extended macro-block (MB), are also implemented in the KTA.;In order to improve the coding efficiency of H.264/AVC, an adaptive prediction block filter (APBF) based on Wiener filter is implemented on every sub-block (SB) of a macro-block (MB) where each MB is decomposed into 4x4 SBs. For each SB using the prediction and reconstruction results of the neighboring SBs the filter coefficients are calculated. The proposed filter is applied to the prediction block of the current SB, and the filtered block is selectively used depending on the rate-distortion (RD) cost. For each SB, if the APBF is used, by reducing the number of bits required for encoding the residual signal between the prediction and original signal of the SB, the coding efficiency is improved. Additionally, as the same filter coefficients can be obtained in the decoder, they do not need to be encoded into the bit-stream. The proposed method achieves a 5.04% bitrate saving on an average when compared to H.264/AVC. |
