Handwritten recognition on Pali cards of Buddhadasa Indapanno

Abstract

This paper proposes a handwritten recognition system on Pali cards of Buddhadasa Indapanno. The proposed system composes of 4 main processes, i.e., image pre-processing, character segmentation, feature extraction, and character recognition. Buddhadasa Indapanno's handwritten images are improved by contrast adjusting, gray scale converting, and noise removing. Then, the characters in the improved images are segmented using connected component labeling and projection profile. The features of each character are extracted by zoning method. After that, these characters are recognized by feedforward back-propagation neural network. The experimental results show that the proposed method yielded the satisfied results.