4-2 機械学習による画素適応予測を併用した可逆符号化方式(第4部門 メディア工学2)

抄録

We previously proposed a lossless image coding scheme based on block-adaptive prediction. In general, prediction performance can be improved if predictor selection is performed pel-by-pel, but this gain would be spoiled by a large amount of additional information needed for the selection. In this paper, we introduce a machine learning technique called support vector machine to enable pel-adaptive prediction without the additional information.