Title :
Feature extraction technique for Hindi numerals
Author :
Sanossian, Hermineh
Author_Institution :
Mutah Univ., Karak, Jordan
fDate :
31 Aug-2 Sep 1998
Abstract :
In this paper a feature extraction technique is presented and applied to printed Hindi numerals. Fourteen different fonts with different sizes are used for training and testing. The method uses local information to extract the features. Classification is performed using neural networks. To reduce the percentage of the error rate a number of networks is trained and a voting system is used to obtain the final result
Keywords :
character recognition; feature extraction; learning (artificial intelligence); neural nets; pattern classification; Hindi numeral recognition; feature extraction; inverse gradient method; learning; neural networks; pattern classification; voting; Data mining; Error analysis; Feature extraction; Image segmentation; Neural networks; Page description languages; Pattern recognition; Real time systems; Testing; Voting;
Conference_Titel :
Neural Networks for Signal Processing VIII, 1998. Proceedings of the 1998 IEEE Signal Processing Society Workshop
Conference_Location :
Cambridge
Print_ISBN :
0-7803-5060-X
DOI :
10.1109/NNSP.1998.710683
