A Fast Mode Decision Algorithm Using Hierarchical and Skip Methods for Intra Prediction in HEVC

This paper proposes an adaptive fast mode decision (AFMD) algorithm for intra prediction in high efficiency video coding (HEVC) using hierarchical and skip methods. AFMD consists of two phases: an adaptive hierarchical mode decision and an adaptive rate-distortion optimization. The proposed AFMD focuses mainly on reducing the number of prediction modes, and thus it can decrease computation time. For rough mode decision and rate-distortion optimization, AFMD computes candidate modes in a way that includes only those intra modes that are most promising. In our experiments, AFMD shows times saving of 41.5% and 40.58% compared to the HEVC test model (HM) reference software, versions 14.0 and 16.0. In addition, the proposed algorithm increases the Bjontegaard delta bitrate by only about 1.00% and 0.93% compared to HM 14.0 and HM 16.0.