エッジ適応型反復法による劣化画像の復元

概要

論文の詳細を見る
This paper proposes a new image restoration method using local characteristics of an image in order to improve its quality. This method restores an image based on a criterion including edge information. We obtain it from the criterion of the conventional Tikhonov-Miller (T-M) regularization method, where the Laplacian operator is replaced by an edge-adaptive operator, which is one of the five operators corresponding to four edge regions with the directions 0,45,90,135 and a flat region. The conjugate gradient algorithm is applied to the above criterion to iteratively restore the image. The over-smoothing effect at an edge, which often occurs in the conventional restoration methods, is prevented by the edge-adaptive operator that smoothes the edge along only one direction. The edge orientation is extracted for each pixel from a temporarily restored image by the conventional T-M method.
社団法人映像情報メディア学会の論文
1997-06-20