7-6 Soft Color-Shrinkage法の最適化のための射影勾配解法(第7部門 画像処理1)

抄録

Previously we proposed the soft color-shrinkage for the sparse coding of a color image. Our previous approach tries to optimize the soft color-shrinkage with a kind of alternately-iterative greedy algorithm, but it achieves the optimization only in the specifically qualified parameter-setting. To cope with this difficulty, this paper constructs a new projected gradient solver, which always achieves the optimization of the soft color-shrinkage irrespective of the parameter setting.