Title :
On Thai character recognition
Author :
De Vel, Olivier ; Wangsuya, Sujint ; Coomans, Danny
Author_Institution :
Dept. of Comput. Sci., James Cook Univ., Townsville, Qld., Australia
Abstract :
Presents a two-stage classification procedure for the recognition of Thai characters. Two classifiers are used to initially classify the characters into different groups and then to classify the characters within each group. Both a backpropagation multi-layered perceptron and a discriminant analysis technique, called penalised discriminant analysis, are used for classification. A comparative performance analysis of the two classifiers is given. To effectively decorrelate the feature vectors, a wavelet transform is also introduced during the preprocessing phase
Keywords :
backpropagation; character recognition; feature extraction; multilayer perceptrons; pattern classification; statistical analysis; Thai character recognition; backpropagation multi-layered perceptron; penalised discriminant analysis; two-stage classification procedure; wavelet transform; Character recognition; Computer science; Data mining; Decorrelation; Feature extraction; Image segmentation; Image sensors; Shape; Statistics; Wavelet transforms;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.488999
