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. |
Year | DOI | Venue |
---|---|---|
1992 | 10.1002/cta.4490200513 | INTERNATIONAL JOURNAL OF CIRCUIT THEORY AND APPLICATIONS |
Field | DocType | Volume |
Character recognition,Feature detection,Pattern recognition,Electronic engineering,Artificial intelligence,Classifier (linguistics),Artificial neural network,Numeral system,Detector,Mathematics | Journal | 20 |
Issue | ISSN | Citations |
5 | 0098-9886 | 4 |
PageRank | References | Authors |
0.65 | 2 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
H. Suzuki | 1 | 238 | 31.31 |
T. Matsumoto | 2 | 4 | 0.65 |
Leon O. Chua | 3 | 1860 | 497.65 |