A fast and low complexity approach for H.264/AVC intra mode decision

  • Authors:

  • Xiuqin Su;Lei Ji;Xiang Li

  • Affiliations:

  • Key Laboratory of Ultrafast Photoelectric Diagnostics Technology, Xi'an Institute of Optics & Precision Mechanics (IOPM), Chinese Academy of Sciences (CAS), Xi'an, China 710119;Key Laboratory of Ultrafast Photoelectric Diagnostics Technology, Xi'an Institute of Optics & Precision Mechanics (IOPM), Chinese Academy of Sciences (CAS), Xi'an, China 710119;Key Laboratory of Ultrafast Photoelectric Diagnostics Technology, Xi'an Institute of Optics & Precision Mechanics (IOPM), Chinese Academy of Sciences (CAS), Xi'an, China 710119

  • Venue:
  • Multimedia Tools and Applications
  • Year:
  • 2011

Quantified Score

Hi-index 0.01

Visualization

Abstract

As one of the main features of H.264/AVC, intra prediction coding technique acts as a basis for encoding performance and efficiency. In official reference software Joint Model (JM), it employs the rate distortion optimization (RDO) technique to get to the best encoding performance. By full searching (FS) all of the candidate modes under the rule of RDO, peak-signal-noise-rate (PSNR) decreases to a very low level, but at the same time, the complexity of calculation increases a lot. Many researchers had devoted to searching the fast algorithm which can decrease the complexity, and had designed so many excellent and intelligent fast algorithms. In this paper we introduced a low complexity and fast approach for H.264/AVC intra prediction algorithm. The new approach is based on reducing the number of candidate modes for further RDO calculation, and decreasing the computational complicacy. It can decide the interpolation direction accurately by calculating the directional pixel-value differences (DPD) of target block, and then do statistic with the obtained values to choose the most probable modes. The experimental results demonstrate that the proposed algorithm can achieve more than 70% time saving than JM, but only a tiny degradation of encoding performance is brought in.