DPCM-Based Edge Prediction for Lossless Screen Content Coding in HEVC

Research output: Contribution to journalArticleResearchpeer-review

11 Citations (Scopus)


Screen content sequences are ubiquitous type of video data in numerous multimedia applications like video conferencing, remote education, and cloud gaming. These sequences are characterized for depicting a mix of computer generated graphics, text, and camera-captured material. Such a mix poses several challenges, as the content usually depicts multiple strong discontinuities, which are hard to encode using current techniques. Differential pulse code modulation (DPCM)-based intra-prediction has shown to improve coding efficiency for these sequences. In this paper we propose sample-based edge and angular prediction (SEAP), a collection of DPCM-based intra-prediction modes to improve lossless coding of screen content. SEAP is aimed at accurately predicting regions depicting not only camera-captured material, but also those depicting strong edges. It incorporates modes that allow selecting the best predictor for each pixel individually based on the characteristics of the causal neighborhood of the target pixel. We incorporate SEAP into HEVC intra-prediction. Evaluation results on various screen content sequences show the advantages of SEAP over other DPCM-based approaches, with bit-rate reductions of up to 19.56% compared to standardized RDPCM. When used in conjunction with the coding tools of the screen content coding extensions, SEAP provides bit-rate reductions of up to 8.63% compared to RDPCM.

Original languageEnglish
Pages (from-to)497-507
Number of pages11
JournalIEEE Journal on Emerging and Selected Topics in Circuits and Systems
Issue number4
Publication statusPublished - 1 Jan 2016


  • Differential pulse code modulation (DPCM)
  • edge prediction
  • high efficiency video coding (HEVC) intra-prediction
  • lossless coding
  • screen content


Dive into the research topics of 'DPCM-Based Edge Prediction for Lossless Screen Content Coding in HEVC'. Together they form a unique fingerprint.

Cite this