PDPモデルによる手書き漢字と平仮名の区別

• Item type: SIG Technical Reports(1)
• 公開日: 1992-01-23
• タイトル: PDPモデルによる手書き漢字と平仮名の区別
• タイトル / 言語: en
• タイトル: Distinction between Hand - written Kanji and Hiragana Characters Using a PDP Model
• 言語: jpn
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: 大阪電気通信大学工学部
• 著者所属(英): en
• 著者所属(英): Faculty of Engineering, Osaka Electro - Communication University
• 著者名: 梅田, 三千雄
• 著者名(英): Michio, Umeda
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: This paper discusses a new distinction system between hand-written Kanji and Hiragana characters using a PDP model. In this system two different types of features are extracted from character patterns. One is a Hough transformed feature which consists of 40 78 or 156 elements. Another is a local direction contributivity (LDC) feature which consists of 64 or 256 elements. A 3-layered artificial neural network based on a parallel distributed processing model is utilized for the distinction. The experimental result shows that the correct distinction rate is above 99% for learned characters data and is about 95% for unlearned characters but same category data. But the rate is less than 90% for unlerned Kanji category data.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: This paper discusses a new distinction system between hand-written Kanji and Hiragana characters using a PDP model. In this system, two different types of features are extracted from character patterns. One is a Hough transformed feature which consists of 40, 78 or 156 elements. Another is a local direction contributivity (LDC) feature which consists of 64 or 256 elements. A 3-layered artificial neural network based on a parallel distributed processing model is utilized for the distinction. The experimental result shows that the correct distinction rate is above 99% for learned characters data and is about 95% for unlearned characters but same category data. But, the rate is less than 90% for unlerned Kanji category data.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA11131797
• 書誌情報: 情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）; 巻 1992, 号 7(1991-CVIM-076), p. 203-210, 発行日 1992-01-23
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会