DMMFast: esquema de redução de complexidade para a predição intra de mapas de profundidade no padrão emergente 3D-HEVC

Abstract

This work presents a complexity reduction scheme for the depth maps intra-frame prediction of the 3D High Efficiency Video Coding (3D-HEVC), called DMMFast (Depth Modeling Modes Fast Prediction). The 3D-HEVC introduces a new set of specific tools for depth map coding, inserting additional complexity to intra prediction, which increases the computational complexity. This increase in complexity results in new challenges in terms of complexity reduction. Therefore, this work presents the DMMFast, a scheme composed of two new algorithms: the Simplified Edge Detector (SED) and the Gradient-Based Mode One Filter (GMOF). The SED classifies the blocks that are likely to be better predicted by the traditional intra modes, avoiding the evaluation of the new modes, called Depth Modeling Modes (DMM), which represents a high computational complexity in the 3D coding. The GMOF is focused on reducing the DMM 1 complexity, the most complexity mode inserted in 3D-HEVC, applying a gradient-based filter in the borders of the block and predicts the best positions to evaluate the DMM 1. Software evaluations showed that DMMFast is capable to achieve a time saving of 24.5% on depth maps prediction, without affecting the quality of the synthesized views. Considering the All Intra setting, the DMMFast is capable to achieve, in average, a time saving of 64% for depth maps coding. Subjective quality assessment was also performed, showing that the proposed technique inserts minimum quality losses in the final encoded video.