ScholarWorks@성균관대학교

초록

In ongoing exploration of video compression beyond VVC capability by JVET in ISO/IEC and ITU-T, intra prediction is being further investigated to utilize more spatial neighbors to predict a current block in many different ways such as directional prediction, non-directional prediction, and offline-trained matrix-based prediction, etc.Noting that the intra prediction under the directional intra mode relies only on the local information, in this paper, we propose a non-local refinement of the predictor generated by the directional intra mode.For this purpose, a block vector of a neighboring block designates a non-local neighboring area for subsequent refinement of the predictor and the refinement is performed on a row and column basis by using predefined weights.Our experiment conducted under the all intra configuration using the first 65 frames of natural video sequences of ECM-11.0 shows 0.02% gain in Y channel, 0.04% and 0.09% loss in Cb and Cr channel. © 2024 SPIE.

키워드