A framework of two-stage combination of multiple recognizers for handwritten numerals

Authors:
K. Lee;Y. Lee
Affiliations:
Electronics and Telecommunications Research Institute, Taejon, Korea;Dept. of Computer Science, Yonsei University, Seoul, Korea
Venue:
PRICAI'00 Proceedings of the 6th Pacific Rim international conference on Artificial intelligence
Year:
2000

Citing 6
Cited 0

Decision Combination in Multiple Classifier Systems

IEEE Transactions on Pattern Analysis and Machine Intelligence
Off-Line Recognition of Totally Unconstrained Handwritten Numerals Using Multilayer Cluster Neural Network

IEEE Transactions on Pattern Analysis and Machine Intelligence
Combination of Multiple Classifiers Using Local Accuracy Estimates

IEEE Transactions on Pattern Analysis and Machine Intelligence
On Combining Classifiers

IEEE Transactions on Pattern Analysis and Machine Intelligence
Multiple Recognizers System Using Two-Stage Combination

ICPR '96 Proceedings of the International Conference on Pattern Recognition (ICPR '96) Volume IV-Volume 7472 - Volume 7472
Recognition of unconstrained handwritten numerals based on dual cooperative neural network

Recognition of unconstrained handwritten numerals based on dual cooperative neural network

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose a framework of two-stage combination method to recognize unconstrained handwritten numerals. It uses multiple combination methods simultaneously unlike the existing methods with only one combination algorithm. The recognizers are first combined by several combination methods at the same time, and the results of them are finally combined by a combination method to generate the final result of recognition. Five recognizers and eight combination methods are used to make a good framework of two-stage combination. The proposed framework was experimented and evaluated with CENPARMI and CEDAR databases. The results showed that we could get the best performance by exploiting the combination methods of different classes at the first stage and then by combining the results of the previous stage by means of Bayesian method.