Fast HEVC intra-prediction mode decision based on conditional selection with hybrid cost ranking

To improve rate-distortion (R-D) performance, high efficiency video coding (HEVC) increases the intra-prediction modes with heavy computational load, and thus the intra coding optimization is highly demanded for real-time applications. According to the conditional probability of most probable mode and the correlation of potential candidate subsets, this paper proposes a fast HEVC intra-prediction mode decision scheme based on conditional selection with hybrid cost ranking. The proposed scheme utilizes the coding results of both rough mode decision and neighboring prediction units to obtain a potential candidate subset, and then conditionally selects the optimal mode through early likelihood decision and hybrid cost ranking. The experimental results demonstrate that the proposed scheme averagely provides about 22.1% encoding speedup with just 0.86% BD-rate loss in comparison with default fast intra prediction in HM16.0. Compared to other fast intra-prediction mode decision schemes, the proposed scheme also significantly reduces intra-coding time while maintaining similar R-D performance in all-intra main profile.