Abstract
Optimization of encoding process in video compression
is an important research problem, especially in the case of
modern, sophisticated compression technologies. In this paper, we
consider HEVC, for which a novel method for selection of the
encoding modes is proposed. By the encoding modes we mean e.g.
coding block structure, prediction types and motion vectors. The
proposed selection is done basing on noise-reduced version of the
input sequence, while the information about the video itself, e.g.
transform coefficients, is coded basing on the unaltered input.
The proposed method involves encoding of two versions of the
input sequence. Further, we show realization proving that the
complexity is only negligibly higher than complexity of a single
encoding. The proposal has been implemented in HEVC reference
software from MPEG and tested experimentally. The results show
that the proposal provides up to 1.5% bitrate reduction while
preserving the same quality of a decoded video.
Go to article