Detection and Reconstruction Scheme for Broken Characters in Thai Character Recognition Systems

• Item type: Journal(1)
• 公開日: 2002-06-15
• タイトル: Detection and Reconstruction Scheme for Broken Characters in Thai Character Recognition Systems
• タイトル / 言語: en
• タイトル: Detection and Reconstruction Scheme for Broken Characters in Thai Character Recognition Systems
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 論文
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_6501
• 資源タイプ: journal article
• その他タイトル / その他のタイトル: 画像・図形認識
• 著者所属: Faculty of Information Technology King Mongkut's Institute of Technology Ladkrabang Bangkok Thailand
• 著者所属: Faculty of Information Technology Dhurakitbundit University Bangkok Thailand
• 著者所属: Faculty of Information Science Mahanakorn Institute of Technology Ba
• 著者所属(英): en
• 著者所属(英): Faculty of Information Technology, King Mongkut's Institute of Technology Ladkrabang, Bangkok, Thailand
• 著者所属(英): en
• 著者所属(英): Faculty of Information Technology, Dhurakitbundit University, Bangkok, Thailand
• 著者所属(英): en
• 著者所属(英): Faculty of Information Science, Mahanakorn Institute of Technology, Bangkok, Thailand
• 著者所属(英): en
• 著者所属(英): Department of Electrical, Electronics and Computer Engineering, School of Science & Engineering, Waseda University
• 著者名: Wichian, Premchaiswadi; Nucharee, Premchaiswadi; Pongsuree, Limmaneewichid; Seinosuke, Narita
• 著者名(英): Wichian, Premchaiswadi; Nucharee, Premchaiswadi; Pongsuree, Limmaneewichid; Seinosuke, Narita
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: This paper proposes an effective scheme for detection and reconstruction of broken printed Thai character images which are one of the major problems in character recognition systems. This problem is still unsolved in commercially available Thai character recognition software packages. The proposed reconstruction method is based on the segment boundaries and distinctive features of Thai characters. This scheme consists of two steps: determination of the number of segments belonging to a character and reconstruction of that particular character image. In the first step it is determined whether each segment is a part of a broken character by using an area of overlap between broken segment boundaries. Then the number of segments belonging to a character image is determined. The reconstruction techniques are based on the number of broken segments. The projection profile an overlapping area and its location are employed to indicate the appropriate areas for connecting broken segments. The test results reveal that the construction is very accurate. The reconstructed characters were also tested by using commercially available Thai character recognition software packages. The test results showed that the proposed scheme significantly improves the recognition rate of the software.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: This paper proposes an effective scheme for detection and reconstruction of broken printed Thai character images, which are one of the major problems in character recognition systems. This problem is still unsolved in commercially available Thai character recognition software packages. The proposed reconstruction method is based on the segment boundaries and distinctive features of Thai characters. This scheme consists of two steps: determination of the number of segments belonging to a character and reconstruction of that particular character image. In the first step, it is determined whether each segment is a part of a broken character by using an area of overlap between broken segment boundaries. Then, the number of segments belonging to a character image is determined. The reconstruction techniques are based on the number of broken segments. The projection profile, an overlapping area, and its location are employed to indicate the appropriate areas for connecting broken segments. The test results reveal that the construction is very accurate. The reconstructed characters were also tested by using commercially available Thai character recognition software packages. The test results showed that the proposed scheme significantly improves the recognition rate of the software.
• 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: AN00116647
• 書誌情報: 情報処理学会論文誌; 巻 43, 号 6, p. 1866-1879, 発行日 2002-06-15
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 1882-7764