Fast SCC in HEVC using a Palette Mode Decision Tree Classifier

Screen Content Coding (SCC) extensions launched as part of the High Efficiency Video Coding (HEVC) standard help in achieving high compression of screen content video with good quality. At the same time the SCC tools come with high computation cost. Such complexity would be a deterrent for the successful application in various real time applications like remote desktop screen sharing, cloud gaming, wireless displays etc. In order to reduce the computational complexity, this paper proposes a decision tree based classifier which classifies Coding Units as Palette or Non Palette blocks. Palette mode computations are skipped for Non palette blocks which results in time saving at the same time with negligible quality degradation. Feature selection and database training for the palette mode decision tree classifier are also discussed in detail. Experimental results show an average time saving of 38.86% with an average BDBR increase of 1.353% and an average BDPSNR degradation of 0.125 dB.

Paper

Full text

PDF

Fast SCC in HEVC using a Palette Mode Decision Tree Classifier

Semantic Scholar · Computer Science · 2020

Abstract

Screen Content Coding (SCC) extensions launched as part of the High Efficiency Video Coding (HEVC) standard help in achieving high compression of screen content video with good quality. At the same time the SCC tools come with high computation cost. Such complexity would be a deterrent for the successful application in various real time applications like remote desktop screen sharing, cloud gaming, wireless displays etc. In order to reduce the computational complexity, this paper proposes a decision tree based classifier which classifies Coding Units as Palette or Non Palette blocks. Palette mode computations are skipped for Non palette blocks which results in time saving at the same time with negligible quality degradation. Feature selection and database training for the palette mode decision tree classifier are also discussed in detail. Experimental results show an average time saving of 38.86% with an average BDBR increase of 1.353% and an average BDPSNR degradation of 0.125 dB.

Similar papers

© 2026 NYSGPT2525 LLC