{"created":"2023-05-15T14:48:42.806588+00:00","id":526,"links":{},"metadata":{"_buckets":{"deposit":"a6089279-df9b-4d2f-84f1-42082b40252b"},"_deposit":{"created_by":3,"id":"526","owners":[3],"pid":{"revision_id":0,"type":"depid","value":"526"},"status":"published"},"_oai":{"id":"oai:kokubunken.repo.nii.ac.jp:00000526","sets":["1:372:373:41"]},"author_link":["16288","16289"],"item_10002_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"1994-03-25","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"20","bibliographicPageEnd":"66","bibliographicPageStart":"23","bibliographic_titles":[{"bibliographic_title":"国文学研究資料館紀要"},{"bibliographic_title":"The Bulletin of The National Institute of Japanese Literature","bibliographic_titleLang":"en"}]}]},"item_10002_description_19":{"attribute_name":"フォーマット","attribute_value_mlt":[{"subitem_description":"pdf","subitem_description_type":"Other"}]},"item_10002_description_5":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"古典原本の電子化の障害となるイメージ雑音除去アルゴリズムの研究を行った。本研究の特色はカラー情報を利用した点にある。研究資料として白色系の和紙に黒色系の墨で書かれた古典原本を用い、イメージ処理として「朱文字」と「裏写り」の除去を試みた。研究成果の概要は以下の通りである。\n\n１）古典原本イメージのカラー情報：古典原本のテキスト・イメージをカラー・イメージスキャナ(100dpi,赤(R)緑(G)青(B)各基本色256諧調）で取り込んだ後、各ピクセルデータをRGB表色系へ展開し、資料のカラー構造を考察した。これより、(a)多くのピクセルは直線R=G=Bの周辺に分布する、(b)朱文字のような特別な色彩のピクセルは(a)とは異なった位置に分布する、(c)RGBの各輝度分布は2峰性を示す傾向があること、が分かった。\n\n2）領域の分割：「和紙」領域と「表文字」領域の分割は、上記(c)の性質より、RGBの各輝度分布に対して「判別閾値選定法」を適用することで達成できた。また(a)及び(b)の性質より「朱文字」領域の分割も可能であった。しかしこの方法では、(a)「表文字」の周辺部が脱落してカスレやすい、(b)「和紙」と「裏写り」の分離が不十分である、という問題点があった。そこで、教師情報なし判別法の一法であるクラスタ分析を適用して分離精度の向上を試みた。本法ではある程度の分離精度の改善を得たが、(a)計算コストが高い、(b)画一的な手法やパラメータの適用では多様な古典原本に対処できない、などの問題点も明確になった。\n\n以上の研究から、古典原本のイメージ雑音処理にカラー情報の利用が有効であることが確認された。\n\n　An image-noise reduction processing algorithm that will be an event in the computerization of the classic original was studied. Characteristic of this study is using color information. As research sources, a classic original written by blackish ink on a whitish washi（和紙）was used, then “letters written in red” and “back copy” were tried to be eliminated. The summaries of research results are as follows.\n\n 1) The color information of classic original image: After taking in text image of a classic original by a color image scanner (100dpi, RGB each basic color 256 gradation), expand each pixel data into Red, Green, Blue color representation system, the color structure of the data was considered. As a result, (a) many pixels are distributed around a straight line R=G=B (b) Pixel of the special color such as red letters are distributed over the different position from (a), (c) it was found that each luminance distribution of RGB tend to be in a two-peak form. \n\n 2) Area division: Division of “Washi” area and “Omote-moji” area were able to achieve to apply “The way to determine and select threshold value” to each luminance distribution of RGB. Division of “red letters” area was also possible by characteristics of (a) and (b). However, there were issues such as (a) Surrounding parts of “Omote-moji” are omitted and easy to be blurred, (b) Insufficient separation of “Washi” and “Back copy”. Therefore a cluster analysis that is one of no instructor information determination method was applied, and improvement of the separation precision was tried. An improvement of separation precision to a certain degree has been made by this method, however, issues became clear such as (a) Calculation cost is high. (b) It is not possible to deal with various classics originals applying a uniform technique and a parameter application.\n\n          It was confirmed that the use of color information is effective for image denoise processing of classics from the above studies.\n","subitem_description_type":"Abstract"}]},"item_10002_identifier_registration":{"attribute_name":"ID登録","attribute_value_mlt":[{"subitem_identifier_reg_text":"10.24619/00000523","subitem_identifier_reg_type":"JaLC"}]},"item_10002_publisher_8":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"国文学研究資料館"}]},"item_10002_source_id_9":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"0387-3447","subitem_source_identifier_type":"ISSN"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"原, 正一郎"},{"creatorName":"ハラ, ショウイチロウ","creatorNameLang":"ja-Kana"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"HARA, SHOICHIRO","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2015-12-10"}],"displaytype":"detail","filename":"KB0158.pdf","filesize":[{"value":"2.8 MB"}],"format":"application/pdf","licensetype":"license_11","mimetype":"application/pdf","url":{"label":"古典原本のイメージノイズ除去に関する一考察","url":"https://kokubunken.repo.nii.ac.jp/record/526/files/KB0158.pdf"},"version_id":"8df2148f-5246-4a63-a0b2-aa9f7d019d52"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"イメージ雑音除去","subitem_subject_scheme":"Other"},{"subitem_subject":"判別閾値選択法と二値化","subitem_subject_scheme":"Other"},{"subitem_subject":"クラスタ分析法と領域分割","subitem_subject_scheme":"Other"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"departmental bulletin paper","resourceuri":"http://purl.org/coar/resource_type/c_6501"}]},"item_title":"古典原本のイメージノイズ除去に関する一考察","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"古典原本のイメージノイズ除去に関する一考察"},{"subitem_title":"Consideration on the Image-noise Reduction Method for Optional Character Recognition","subitem_title_language":"en"}]},"item_type_id":"10002","owner":"3","path":["41"],"pubdate":{"attribute_name":"公開日","attribute_value":"2014-11-07"},"publish_date":"2014-11-07","publish_status":"0","recid":"526","relation_version_is_last":true,"title":["古典原本のイメージノイズ除去に関する一考察"],"weko_creator_id":"3","weko_shared_id":-1},"updated":"2023-05-16T02:40:30.406561+00:00"}