Portal de Periódicos da CAPES

A New Search Engine for Chinese Document Image Retrieval Based on Chinese Character Segmentation Features

2002; World Scientific; Volume: 15; Issue: 04 Linguagem: Inglês

10.1142/s0219427902000765

2010-0205

Tung-Shou Chen, Ming-Chiang Chen,

Handwritten Text Recognition Techniques

International Journal of Computer Processing of LanguagesVol. 15, No. 04, pp. 417-431 (2002) Special Issue: Web and WAP Oriental Languages Multimedia Computing; Guest Editors: Timothy Shih and Jianhua MaNo AccessA New Search Engine for Chinese Document Image Retrieval Based on Chinese Character Segmentation FeaturesTUNG-SHOU CHEN and MING-CHIANG CHENTUNG-SHOU CHENDepartment of Information Management, National Taichung Institute of Technology, Taichung, Taiwan, Republic of China Search for more papers by this author and MING-CHIANG CHENDepartment of Information Management, National Taichung Institute of Technology, Taichung, Taiwan, Republic of China Search for more papers by this author https://doi.org/10.1142/S0219427902000765Cited by:2 PreviousNext AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack CitationsRecommend to Library ShareShare onFacebookTwitterLinked InRedditEmail AbstractDocument images have been widely used on the Internet and in digital libraries. Traditionally, the retrieval systems have been used by keyword searching via OCR technology. In this paper, we propose a new image retrieval system based on the segmentation features of characters for Chinese document images, so users can retrieve the document image directly from a database without the OCR phase. From an image viewpoint, a Chinese character consists of many segments. For example, could be separated into two segments. One is, and the other is. So, the number of segments of each Chinese character image is used as its corresponding codes, and the list of the codes is applied to denote the document image. Users input a document image into the proposed system first, and then the system matches the correct image from the database by using the codes. According to the experimental results, the search of the proposed system is efficient and effective. Furthermore, the proposed system satisfies the image retrieval requirement. It is a Chinese document image retrieval based on segmentation.Keywords:Document ImageChinese Character SegmentationImage Duplicates FiguresReferencesRelatedDetailsCited By 2Self-Adaptability Research of Billet Character Segmentation in the Complex SceneYang Yijun, Hong Hanyu, Wang Yiwen and Zhang Xiuhua1 Jul 2012A new algorithm of mixed Chinese-English character segmentation based on irregularty degree Jian-Hui Si, Fang Yang and Xue-Dong Tian1 Jul 2008 Recommended Vol. 15, No. 04 Metrics History KeywordsDocument ImageChinese Character SegmentationImage DuplicatesPDF download