A pattern recognition approach for handwritten hiragana character identification

When it is about huge document retrieval, searching for a particular area for handwritten text recognition becomes a difficult job to do. It is very important for identification architecture to have proper linguistic knowledge for recognizing a character from an image document. This paper proposes a method of character identification for handwritten Japanese text using features of the handwritten pictorial character. It follows up with an analytical discussion of the pattern and characteristic features for hand written Japanese hiragana scripts, based on which the approach have been proposed. Experimental studies on handwritten samples show how the approach could effectively improve the performance of the recognition system.