An Efficient Table Prediction Scheme for CAVLC in H.264

The most recent video coding standard H.264 has adopted context-based adaptive variable-length coding (CAVLC) as the entropy coding tool in the baseline profile. By combining an adaptive variable-length coding technique with context modeling, we can achieve a high degree of redundancy reduction. However, CAVLC in H.264 has a weakness that the correct prediction rate of the variable-length coding (VLC) table is low. In this paper, we propose a new VLC table prediction scheme considering multiple reference blocks; the block at the same position in the previous frame and neighboring blocks of the current block in the current frame, and weighting values considering quantization parameter (QP). Based on the correctness of the VLC table prediction depending on QP values for each reference block, we determine weighting values. Experimental results show that the proposed algorithm increases the correct prediction rate of the VLC table by 7.18 % and reduces bit rates by 0.78 %, compared to the H.264 algorithm.