HEVC intra mode selection based on Rate Distortion (RD) cost and Sum of Absolute Difference (SAD)

Junaid TARIQ, Sam KWONG, Hui YUAN

Research output: Journal PublicationsJournal Article (refereed)peer-review

34 Citations (Scopus)

Abstract

High Efficiency Video Coding (HEVC) encoder provides higher compression efficiency by offering 35 intra modes. However, the encoding complexity is increased due to more modes are involved in the decision process. Therefore, it is desired to build a fast intra prediction algorithm that is practical for real time application. In this paper, a quadratic approach for reducing intra coding complexity is proposed. Firstly, the relationship between the RD-cost and the SAD is investigated. Secondly, a model is proposed to estimate the RD-cost of all 35 intra modes using the quadratic relation, thus avoiding the computation of entropy coding, Hadamard cost, distortion, and transform. Experimental results demonstrate that the average time saving of the proposed approach is 31-38%, while the BD-Bit Rate increment is only 0.62-1.37%, respectively.
Original languageEnglish
Pages (from-to)112-119
JournalJournal of Visual Communication and Image Representation
Volume35
DOIs
Publication statusPublished - 1 Feb 2016
Externally publishedYes

Funding

This work is supported in part by City University of Hong Kong Strategic Grant 7004418, and in part by the National Natural Science Foundation of China under Grants 61272289, 61571274; in part by the City University of Hong Kong Shenzhen Research Institute, Shenzhen, China; and in part by the Young Scholars Program of Shandong University (YSPSDU) under Grant 2015WLJH39.

Keywords

  • Fast intra prediction
  • High Efficiency Video Coding (HEVC)
  • Image coding
  • Low complexity video coding
  • Rate distortion cost prediction
  • Rough mode decision
  • Video coding
  • Video compression

Fingerprint

Dive into the research topics of 'HEVC intra mode selection based on Rate Distortion (RD) cost and Sum of Absolute Difference (SAD)'. Together they form a unique fingerprint.

Cite this