Content Adaptive Intra Prediction Algorithm for HEVC Encoder

Abstract

The emerging high efficiency video coding (HEVC) standard is better than the previous video coding standard H.264/AVC on coding efficiency and provides equivalent quality with 50% bit-rate reduction. In light of intra coding, HEVC reduces data amount by around 36% on average compared with H.264/AVC. The benefit of these improvements is counterbalanced by computational complexity. To reduce the computational load, this paper proposes a content adaptive intra prediction algorithm for coding unit (CU) depth and prediction unit (PU) mode decisions based on spatial correlation and content information for the HEVC encoder. The proposed scheme utilizes the correlation among neighboring CUs and analyzes the edge information by applying the classification in MPEG-7 to abate the number of candidates of CU and PU modes. Moreover, the proposed algorithm presents a simplified rate-distortion cost (SRDO) by using the subsampled sum of absolute difference (SSAD) to curtail the computational burden. Experimental results show that our algorithm achieves a 34.28% time saving with only a 0.68% BD-rate increase on average for all intra configuration compared to HM11.0, and performs better than previous studies.