抽出された特徴による手書き漢字構造の再生
スポンサーリンク
概要
- 論文の詳細を見る
There is very little investigation on the process of an pattern being normalized by making use of the topologically invariant features extracted from it. In this paper, attention will be paid mainly to that a kind of automatically input pattern-normalizing process may be explained by the faculty of reproducing the corresponding input (which is not structured) by making good use of a ready-made mean information pattern in co-operation with the topologically invariant features extracted from the input pattern, where the reproduced pattern is called a structured information pattern. That is, this paper mainly discusses a method of reproducing an input pattern so that the reproduced pattern may be correctly recognized, which is provided with the faculty of automatically normalizing the corresponding input pattern by putting the topologically invariant features to use. In addition, the corresponding reproduced patterns (which remain invariant over an expaision-and-contraction transformation group) of hand-written Chinese characters are determined with a computer simulation and the related explanation is given in detail.
- 一般社団法人情報処理学会の論文
- 1977-11-15
著者
関連論文
- 認識主体の集合のなすハウスドルフ位相群
- 5p-FA-12 第一量子化の効果を含んだ第二量子化(第三量子化)
- 30a-D-6 波動関数の構造化, 情報量, 物理量のHamiltonianとの非可換性
- 5p-J-13 波動関数の近似表現
- 13a-E-2 認識の量子論
- 抽出された特徴による手書き漢字構造の再生
- 手書き漢字の側抑制効果的分解とその計算機シミュレーション
- 標本化音声記号から抽出される測度的不変量
- 発見的探索形位相不変認識システム
- 認識の主要な局面と認識諸定理
- 特徴量としての測度的ウニタリ不変量の完全な集合の一構成(技術談話室)
- 構造化情報パタ-ンの4性質(技術談話室)
- 感覚空間回路の設計と作用素に対するラプラス変換法