earticle

논문검색

A Fast Decision Algorithm for High Efficiency Video Coding

초록

영어

Fast mode decision algorithms in video intra coding are very useful and necessary for implementing practical real-time performance of the encoder. It can reduce the computational complexity by excluding impossible modes from candidate set. Anyway, most of these algorithms may give rise to the remarkable quality degradation, especially for the all intra coding scenario. In order to improve the encoding speed while avoiding the quality loss, this paper presents a new algorithm for fast CU size decision and mode decision in HEVC intra coding. First, the size of the coding unit is pre-determined according to judging the coding unit whether has the sub-block whose depth is 3 and analyzing the depth of the adjacent CU. Then a judgment by block-DCT coefficients is made on the texture direction of image block, determining candidate mode set based on the calculated direction and the best prediction mode of the adjacent block, which can reduce the number of candidate mode, thereby reducing the amount of computation to save coding time. The experimental results show that the algorithm provides about 34.6% reduction of intra coding time compared with the latest HM13.0 with negligible loss of PSNR and a little increase of bite-rate.

목차

Abstract
 1. Introduction
  1.1 Fast Coding Unit Size Decision
  1.2 Fast Intra Prediction Mode Decision
 2. Overview of Intra Prediction in HEVC
 3. The Proposed Fast Decision Algorithm
  3.1 Fast CU Size Decision Based on the Adjacent Code Block
  3.2 Fast Mode Decision Based on the Calculation of the Texture Angle
  3.3 Algorithm Description
 5. Conclusions
 References

저자정보

  • Yong Gan School of Computer and Communication Engineering, Zhengzhou University of Light Industry, 450002, Zhengzhou, China
  • Xiaorong Zhao School of Computer and Communication Engineering, Zhengzhou University of Light Industry, 450002, Zhengzhou, China
  • Tianbao Li School of Computer and Communication Engineering, Zhengzhou University of Light Industry, 450002, Zhengzhou, China

참고문헌

자료제공 : 네이버학술정보

    함께 이용한 논문

      ※ 원문제공기관과의 협약기간이 종료되어 열람이 제한될 수 있습니다.

      0개의 논문이 장바구니에 담겼습니다.