ESTG - Comunicações em conferências e congressos internacionais
Permanent URI for this collection
Browse
Browsing ESTG - Comunicações em conferências e congressos internacionais by Author "Agostini, Luciano"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- Classification-based early termination for coding tree structure decision in HEVCPublication . Correa, Guilherme; Assunção, Pedro; Cruz, Luis A. da Silva; Agostini, LucianoThe High Efficiency Video Coding (HEVC) standard provides improved compression rates in comparison to its predecessors at the cost of large increases in computational complexity. An important share of such increases is due to the introduction of flexible Coding Tree structures, which best configuration is decided through exhaustive tests in a Rate-Distortion Optimization (RDO) scheme. In this work, an early termination method for the decision of such structures was designed using classification trees obtained through Data Mining techniques. The classification trees were trained using intermediate encoding results from a training set of video sequences and implemented in the encoder to skip the full RDO-based decision. An average reduction of 37% in the HEVC encoder computational complexity was achieved when using the designed classification trees, with a negligible cost of only 0.28% in terms of Bjontegaard Delta-rate increase.
- Four-step algorithm for early termination in HEVC inter-frame prediction based on decision treesPublication . Correa, Guilherme; Assunção, Pedro; Agostini, Luciano; Cruz, Luis A. da SilvaThe flexible encoding structures of High Efficiency Video Coding (HEVC) are the main responsible for the improvements of the standard in terms of compression efficiency in comparison to its predecessors. However, the flexibility provided by these structures is accompanied by high levels of computational complexity, since more options are considered in a Rate-Distortion (R-D) optimization scheme. In this paper, we propose a four-step early-termination method, which decides whether the inter mode decision should be halted without testing all possibilities. The method employs a set of decision trees, which are trained offline once, using information from unconstrained HEVC encoding runs. The resulting trees present a mode decision accuracy ranging from 97.6% to 99.4% with a negligible computational overhead. The method is capable of achieving an average computational complexity decrease of 49% at the cost of a very small Bjontegaard Delta (BD)-rate increase (0.58%).