論文 ·日本語 ·未確認
A CNN handwritten character recognizer
Hirotaka Suzuki ・ Tadashi Matsumoto ・ Leon O. Chua
- 刊行年
- 1992-09-01
- 収録
- 『International Journal of Circuit Theory and Applications』 20(5) pp. 601-612
- 出版
- Wiley
- 言語
- 英語
- OpenAlex
- W1981795695
- DOI
- 10.1002/cta.4490200513
- MAG
- 1981795695
- ISSN
- 0098-9886
- URL
- https://doi.org/10.1002/cta.4490200513
要旨
Abstract CNNs are used for feature detection in handwritten character recognition. Detected features are fed to a simple classifier network. Performance was tested by using two well‐known ETL data base series: (i) ETL3 consisting of numerals, alphabets and several symbols and (ii) ETL8B2 consisting of Japanese Hirakana characters. the average recognition rate for ETL3 is 94.8%, while that for ETL8B2 is 85.7%. Both series include ‘hard’ characters so distorted that even humans cannot recognize them.
主題
この書誌の出所
- openalex— W1981795695（2026-08-14取得）
引用
Hirotaka Suzuki・Tadashi Matsumoto・Leon O. Chua（1992-09-01） A CNN handwritten character recognizer 『International Journal of Circuit Theory and Applications』 20(5) pp. 601-612 Wiley
Suzuki1992CNNHandwrittenCharacter