A Fast Intra Mode Prediction Decision Algorithm Based on Neighborhood Grouping for HEVC

Abstract

Abstract High Efficiency Video Coding (HEVC) employs quadtree coding tree unit (CTU) structure to improve its coding efficiency, but at the same time, it requires a very high computational complexity due to its exhaustive calculating process to find the optimal prediction mode for the current CU (Coding Unit). Aiming to solve the problem, a fast intra mode prediction decision algorithm based on neighborhood grouping is presented for HEVC in this article. The contribution of this article lies in the fact that the authors successfully use the neighborhood grouping technology to rapidly find the optimal prediction mode for the current CU, which can significantly reduce computation complexity for HEVC. Specifically, they use the correlative information of adjacent angle prediction mode in their first scheme to quickly reduce the number of Rough Mode Decision (RMD); then they use the correlation between Most Probable Mode (MPM) and the first candidate mode selected as the optimal prediction mode to quickly determine the number of candidate modes; at last, they quickly find the optimal prediction mode with the minimum Rate Distortion Cost (RDCost) by using neighborhood grouping technology or direct calculation approach based on the number of candidate modes. As a result, their proposed algorithm can efficiently solve the problem above. The simulation results show that our proposed intra mode prediction decision algorithm based on neighborhood grouping in this article can reduce about 22.3% computational complexity on average only at a cost of 0.03% bit rate increase and 0.26% PSNR decline compared with the standard reference HM16.1 algorithm.