A CNN handwritten character recognizer
- 1 September 1992
- journal article
- Published by Wiley in International Journal of Circuit Theory and Applications
- Vol. 20 (5) , 601-612
- https://doi.org/10.1002/cta.4490200513
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.Keywords
This publication has 7 references indexed in Scilit:
- A CNN chip for connected component detectionIEEE Transactions on Circuits and Systems, 1991
- CNN cloning template: connected component detectorIEEE Transactions on Circuits and Systems, 1990
- CNN cloning template: shadow detectorIEEE Transactions on Circuits and Systems, 1990
- VLSI implementation of a neural network modelComputer, 1988
- Neocognitron: A neural network model for a mechanism of visual pattern recognitionIEEE Transactions on Systems, Man, and Cybernetics, 1983
- Distinctive features in automatic recognition of handprinted charactersSignal Processing, 1982
- Feature extraction: A surveyProceedings of the IEEE, 1969