Experiments are conducted using ten fold cross validation and different learning rates and momentum. The accuracy obtained is comparable with the results obtained from other researchers in literature. The proposed method is extremely fast in both feature extraction and classification. Further work needs to be done to improve the classification accuracy.