Achieving high recognition accuracy on historical handwritten characters is a challenging problem, for which many different solutions have been published in the literature till date. For the Historical handwritten Korean documents (Min et al; 2004) proposed a recognition based character segmentation. LDA classifier was used to perform classification based on geometric features and context information. 92.98% recognition rate was reported. (Vamvakas et al; 2008) proposed a classification system based on SVM and RBF kernel for Historical documents. The system was trained on pixel density of character zones and could achieve an accuracy of 95%. In (G Vamvakas; 2009) features based on mass of each sub image of the characters were extracted and fed to a hierarchical classifier to recognize Greek Historical Documents with an accuracy of 94%. A decision tree approach based on correlation coefficient character grouping was employed by (Sastry et al; 2010) for classification of Telugu characters extracted from palm leaves with an accuracy of 93%. (Raj Kumar et al; 2012) proposed a self-adaptive learning SVM classifier based on time domain and frequency domain features. The method achieved 94% accuracy in recognition of 8th century Tamil consonants.